Intelligent management and control method and system for energy-saving illumination of highway tunnel and computer program

Abstract

The invention relates to the technical field of intelligent traffic and energy-saving illumination, in particular to an intelligent management and control method and system for energy-saving illumination of a highway tunnel and a computer program, and is suitable for an intelligent illumination control system for improving tunnel driving safety and energy efficiency of an illumination system. Dynamic response and fine energy consumption adjustment of tunnel lighting are realized through out-of-tunnel brightness multi-source fusion prediction, entrance section feedforward, closed-loop dimming control, a segmented intelligent lighting strategy based on vehicle detection and an in-tunnel brightness closed-loop and stepless dimming mechanism. The system has a multi-sensor redundancy mechanism and can automatically return to a safety mode when equipment fails, so that the continuity and the safety of illumination are guaranteed; and meanwhile, data summarization and energy-saving statistics are combined, control parameters are optimized in cooperation with a digital twinning or self-learning algorithm, the energy-saving effect and the operation and maintenance efficiency are further improved, and the purposes of traffic safety and low-carbon operation are considered.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and energy-saving lighting technology, specifically to an intelligent control method, system and computer program for energy-saving lighting in highway tunnels, which is applicable to an intelligent lighting control system that improves tunnel driving safety and the energy efficiency of the lighting system. Background Technology

[0002] With the sustained and rapid development of my country's economy and society and the continuous improvement of transportation infrastructure, highways and tunnels, as important components of modern road transportation, play an irreplaceable role in alleviating traffic congestion, improving transportation efficiency, and ensuring driving safety. In recent years, with the growing popularity of energy conservation, emission reduction, and low-carbon economy concepts, highway tunnel lighting systems, which account for a significant portion of tunnel energy consumption, have received increasing attention for their energy-saving technologies and intelligent management. Especially against the backdrop of the rapid development of LED lighting and next-generation information technologies such as the Internet of Things, big data, and artificial intelligence, upgrading traditional tunnel lighting methods to achieve energy conservation, emission reduction, and improved operational efficiency has become a pressing technical challenge.

[0003] Traditional highway tunnel lighting systems rely primarily on manually set timings and simple timer switches for control, typically operating in a continuous, full-power mode. This makes it difficult to flexibly adjust lighting based on real-time traffic flow and changes in the external environment. Especially during periods of low traffic or when no vehicles are passing through, the tunnel lights remain fully on, resulting in significant energy waste. Furthermore, sudden changes in ambient light when vehicles enter or exit the tunnel can easily cause visual shocks for drivers, such as the "black hole effect" or "white hole effect," thus affecting driving safety. While traditional tunnel lighting technology has begun to adopt LED energy-saving lamps to replace traditional light sources in recent years, most systems remain at the stage of timed control or brightness adjustment based solely on a single sensor, failing to achieve intelligent management that automatically adjusts lighting brightness according to real-time environmental and traffic conditions.

[0004] Traditional tunnel lighting systems have the following main drawbacks: 1. Low level of intelligence and simplistic dimming strategies: Currently, most tunnel lighting systems use pre-set timed switching modes, lacking the perception and feedback of dynamic variables such as real-time traffic volume and ambient light. Even when some automatic dimming technologies are used, they mainly rely on data from a single sensor (such as relying solely on ambient light detection values ​​at the entrance), failing to fully integrate various information. Therefore, when traffic volume is low or there are no vehicles, the tunnel lighting remains fully bright, resulting in significant energy waste; and when the light suddenly changes, the system's adjustment is not timely, which may cause visual adaptation difficulties for drivers and reduce driving safety.

[0005] 2. Insufficient data collection and fusion: In the existing system, the detection of the lighting environment outside and inside the tunnel mostly relies on single - type sensors, such as traditional brightness detectors or illuminometers. The measurement data is easily affected by installation location, external occlusion, and environmental factors (such as rain, snow, fog, etc.), resulting in poor stability. At the same time, the lighting requirements in different areas (such as the entrance, transition, middle section, and exit section) are different, and it is difficult for single - collected data to meet the refined regulation requirements of each area. In addition, the vehicle detection system usually only relies on video surveillance or radar, lacking vehicle - road cooperation and multi - source redundancy technology support, so detection errors are likely to occur under complex traffic conditions.

[0006] 3. Imperfect light decay compensation and dimming closed - loop control: Although LED lighting technology has advantages such as high efficiency, energy saving, and long lifespan, with the extension of usage time, problems such as light decay, brightness reduction, and color change exist. Currently, some systems fail to monitor and compensate for the light decay of lamps in real time, resulting in a deviation between the actual output brightness and the design parameters, which in turn affects the lighting uniformity and safety inside the tunnel. Existing closed - loop dimming control generally only relies on the brightness measurement data at the current moment, lacking intelligent adjustment based on historical data and feed - forward prediction, and unable to quickly respond to the influence of ambient light changes and lamp attenuation.

[0007] 4. Poor system robustness and insufficient fault tolerance: Traditional systems rely heavily on sensors and communication networks. Once a sensor fails or a communication fault occurs, the entire dimming system is prone to getting out of control, unable to adjust the lighting mode in a timely manner, which poses a risk to driving safety. At the same time, in case of emergency events such as fires and vehicle accidents, the system fails to achieve global linkage dimming and cannot meet the emergency lighting requirements.

[0008] Chinese invention patent (Publication number: CN117412442A, Publication date: January 16, 2024) discloses an energy - saving lighting control behavior that accurately locates the lighting area to be controlled by generating lighting conversion information based on the monitoring data of the target lighting area and the urban tunnel, and can perform energy - saving lighting control according to the requirements of the actual lighting scene to minimize energy consumption and achieve energy - saving effects. However, this solution mainly focuses on the scene feature modeling of urban environments and complex sections, and is not fully applicable in long - distance highway tunnels and energy - saving modes mainly based on vehicle - following dimming. There is still a large room for improvement, especially in aspects such as vehicle prediction, segmented dimming, brightness closed - loop control, emergency mechanisms, and energy - saving statistics optimization. Summary of the Invention

[0009] To address the aforementioned issues, this invention proposes an intelligent control method for energy-saving lighting in highway tunnels. This method can accurately calculate the target brightness at the entrance section and dynamically adjust the lighting brightness as needed inside the tunnel, as well as achieve real-time closed-loop control of the tunnel's brightness. This ensures that tunnel lighting meets both traffic safety requirements and achieves high energy efficiency and extends the lifespan of the lamps.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: A method for intelligent management and control of energy-saving lighting in highway tunnels includes the following steps: 1) External brightness acquisition and entrance dimming: The external ambient brightness data at the tunnel entrance is acquired using multi-source sensing devices, including brightness detectors, video image recognition, and optional weather and sunshine forecast information. The multi-source observations are then time-series aligned and fused to obtain the real-time external tunnel brightness. L out (t) and short-term forecasts The fusion results are combined with tunnel lighting design specifications to obtain the target brightness of the entrance section. When detected When a rapid increase or decrease is imminent, the lighting power of the entrance section is adjusted in advance using a feedforward + closed-loop dimming algorithm to avoid the black hole effect or over-illumination. 2) Vehicle inspection and segmented control: Vehicle information is collected at the tunnel entrance; the tunnel interior lights are divided into several independent dimming zones at preset intervals, and vehicle detectors and dimming controllers are deployed in each dimming zone; when a vehicle is detected about to enter a dimming zone, that zone is switched from standby brightness in advance. B standby Adjust to vehicle brightness B car (z,t) After the vehicle leaves and it is confirmed that no new vehicle is left behind, the brightness of the section will be restored to standby mode. 3) Closed-loop brightness control and stepless dimming within the cave: Brightness detectors were installed at the entrance, transition, middle, and exit sections to measure the actual brightness of each section in real time. The dimming coefficient was corrected based on the luminous decay model of the lamps, and combined with the brightness outside the cave. Lout (t) or traffic flow information, using a stepless dimming algorithm to perform feedforward + closed-loop control on the output power of each segment, so that the actual brightness meets the target brightness required by the specification. ; 4) Abnormal state handling and redundancy mechanisms: When the vehicle detection sensor malfunctions or goes offline, it automatically reverts to stepless dimming or sequential dimming mode to maintain basic lighting; when the brightness detector malfunctions or goes offline, it automatically switches to the preset sequential dimming mode; when a fire or other emergency event is detected, the emergency lighting mode is triggered, switching the relevant section or the entire tunnel to the highest brightness.

[0011] Preferably, the multi-source fusion and prediction in step 1) includes: 1.1) The following observation sources shall be deployed outside the tunnel entrance: a dedicated brightness detector at a fixed angle shall be installed at a certain distance outside the entrance to periodically output the measured values. The environment at the tunnel entrance is captured by a camera, and an image brightness recognition algorithm is used to obtain... Obtain solar radiation forecasts for the next 1-3 hours from the meteorological platform API. Or cloud cover rate, rainfall probability index; combine the above data at time t to form the original observation: ; 1.2) Through the fusion function f fusion Calculate real-time external brightness L out (t); 1.3) Employing a time-series forecasting model f pred right L out (t) is used to make short-term predictions and obtain This is used to perform feedforward dimming on the entrance section lights; The inlet dimming in step 1) adopts the following mathematical model: Set entrance section dimming command u entr (t)∈[0,1]; Feedforward term: ; Closed-loop term: , ; Dimming command limit: , And mutations can be suppressed through a smoothing function.

[0012] Preferably, step 2) includes the following steps: 2.1) Multi-source hyperfusion detection Multiple sensors are deployed at the tunnel entrance and inside the tunnel, including: AI video cameras, millimeter-wave or lidar, geomagnetic sensors, and roadside units that interact with the vehicle's onboard unit; deep data fusion algorithms are used to perform unified time alignment of information from different sensors and the vehicle's own information to generate an initial record of the vehicle's status; 2.2) Segmented dimming and distributed control The tunnel is divided into several dimming sections, and each section is equipped with a vehicle detector and dimming controller integrated device with local intelligent functions. When a vehicle enters or is about to enter the section, the brightness is increased. After leaving, if there is no new vehicle after a certain delay, the brightness is restored to standby mode. The lighting transition between sections is achieved through neighborhood collaboration. 2.3) Traffic Flow Prediction and Adaptive Parameter Tuning The backend or edge server uses a time-series prediction model to predict short-term traffic flow based on historical traffic flow, weather, and holiday information. Then, it automatically adjusts the advance lighting distance and delay of the dimming time for each section based on the prediction results. When the actual speed of the vehicle deviates from the design speed, it dynamically changes the lighting triggering time of the next section to avoid insufficient or wasted lighting due to speeding or low speed. 2.4) Self-learning and digital twins In the background, reinforcement learning or evolutionary algorithm analysis is performed on the tunnel operation data to optimize the control parameters of segmented dimming; a digital twin simulation environment is established to simulate and evaluate the new strategy; if the safety and energy-saving indicators are good, the strategy is then implemented in each section of the tunnel.

[0013] Preferably, in step 2): 1) Define the tunnel length [0, L] total The area is divided into N dimming zones, each equipped with a vehicle detector and a dimming controller. 2) When a vehicle approaches the section boundary in advance, the system will switch the corresponding section from standby brightness to vehicle-enabled brightness. 3) After the vehicle leaves, if the section is within the delay time t delay If no new car enters, the brightness will revert to standby mode (B). standby ; 4) Adjacent sections can exchange detection data to prevent excessive brightness jumps.

[0014] Preferably, in step 2): 1) For vehicles equipped with roadside units (OBUs), data is acquired before they enter the tunnel. ; ID i For license plate or OBU identification; t entry This refers to the estimated time to enter the tunnel; v est、 a estThese are the vehicle's estimated speed and acceleration, respectively, from its built-in sensors. 2) If the vehicle does not have a roadside unit, the initial observation Z is obtained through AI video at the entrance, radar, and geomagnetic sensors. i At time t k Data association is performed internally; 3) Predict using uniform acceleration or uniform velocity models And update the vehicle state using minimum distance matching or a neural network correlator when new data is observed. X i ( t k ); 4) Segment mapping rules: If x i (t)∈[S z−1 If the vehicle is in section z and the dimming controller of section z is responsible for the dynamic dimming command, then the vehicle is in section z.

[0015] Preferably, step 3) includes the following steps: High-precision brightness detectors are deployed at at least four key sections within the tunnel: entrance E, transition T, basic section B, and exit X, to collect real-time actual brightness data. ; 3.1) The brightness state and light decay model are as follows: remember For a moment t k The road surface brightness values ​​measured in section z, k=0,1,2,… remember The light decay ratio of the luminaire in segment z is estimated from a prior light decay model or detection data: , F is an empirical formula or a function of cumulative running time, temperature, and dust deposition factors; When the controller issues a relative brightness command At that time, the nominal output brightness is set to After considering light decay, the theoretical brightness reaching the road surface is: ; 3.2) Reference brightness and external environment remember This represents the real-time measured brightness outside the tunnel; according to the "Detailed Rules for Lighting Design of Highway Tunnels," a minimum target brightness is defined for each section z. Common situations: Entry section: ; Transition section: Decreasing function +Standardized fixed coefficients; Middle section: ≥ It can be finely adjusted according to traffic flow; Export section: ≥ And take into account the influence of external brightness; Overall reference brightness: , Where L safe It is a safety baseline value to ensure that the minimum video surveillance and emergency response requirements are met; 3.3) Stepless dimming control model make e z (t k ) Indicates the brightness error of segment z: , in, That is, the actual measured brightness; To ensure fast and stable dimming, a combination of feedforward and closed-loop methods is used: Feedforward term: Based on the estimated light decay coefficient and nominal brightness curve, first calculate the following: , Where Γ = is a calibration constant relating the rated maximum brightness of the luminaire to the dimming signal; Closed-loop correction term: using a PI to adjust the error e z (t k ) Make corrections; PI method: ; Where K p ,K i For proportional or integral gain, adaptive or online scheduling; Output limiting:

[0016] To avoid frequent fluctuations in dimming commands, perform index smoothing or speed limiting operations: .

[0017] Preferably, in step 4): When the vehicle detector fails to output data for ΔN consecutive times or the radar goes offline, it automatically reverts to stepless dimming mode and maintains the safe lower limit of tunnel illumination. When the brightness detector malfunctions, the system switches to sequential dimming or fixed power distribution scheduled by seasonal time periods. If a fire, accident, or emergency triggers E mergency When =true, max{B} is executed on the relevant section or the entire tunnel. z (t)}=1.0 is the highest brightness command.

[0018] Preferably, the method also includes uploading the time-series data collected by vehicle detection, brightness detection, and dimming controller to the background; comparing the energy consumption of the constant-on mode, calculating the energy saving rate and outputting an energy saving report; adaptively optimizing the dimming parameters based on the statistical results, and improving the overall lighting efficiency by combining digital twin or self-learning algorithms. And / or, the method also includes manual intervention and management: providing manual dimming and one-click full brightness functions on the monitoring platform to deal with sudden weather or emergencies; when maintenance personnel intervene, any segment can be temporarily switched to the highest brightness or restored to automatic control.

[0019] As a preferred approach, energy consumption is compared based on the following formula: , Where P z B represents the full power of the lamp in segment z. z (t)∈[0,1] is the dimming coefficient, E full The total energy consumption when all sections are constantly lit at full power is recorded, and the energy saving rate η is recorded to guide the adaptive optimization of subsequent dimming strategies.

[0020] Furthermore, the present invention also discloses an intelligent control system for energy-saving lighting in highway tunnels, used to implement the method, comprising: The external brightness acquisition module, a multi-source sensing device deployed outside the tunnel entrance, is used to collect and fuse data to output L. · (t) and ; The vehicle detection and segmented control module includes sensors and dimming controllers deployed at the tunnel entrance and in each independent dimming section, used to output segmented lighting / darkening commands based on vehicle trajectory information; The tunnel brightness closed-loop and stepless dimming module is connected to the brightness detector of the key section of the tunnel. It is used to perform light attenuation compensation, feedforward + closed-loop dimming algorithm, and send dimming commands to the corresponding lamps. The abnormal state handling and redundancy mechanism module automatically performs timing dimming or maximum brightness when the detector malfunctions or an emergency event is triggered. The data aggregation and energy-saving statistics module is used to aggregate vehicle and brightness data, generate energy-saving reports, and adaptively iterate dimming strategies in conjunction with big data analysis. The manual intervention and management interface provides human-computer interaction functions for the monitoring platform, allowing maintenance personnel to manually turn on all lights, make temporary corrections, or switch back to automatic control.

[0021] Furthermore, the present invention also discloses a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the method.

[0022] Furthermore, the present invention also discloses a computer program product, including a computer program or instructions that, when executed by a processor, implement the method.

[0023] By employing the aforementioned technical solutions, the present invention provides an intelligent management and control method and system for energy-saving lighting in highway tunnels. Through the comprehensive application of multiple key technologies, including multi-source information fusion, predictive feedforward, closed-loop adaptive dimming, regionalized segmented control, and fault redundancy, the present invention achieves the following significant technical effects: 1. High-precision dynamic environmental perception and predictive response: This invention employs multiple sensors, including brightness detectors, video image recognition, and weather forecasts, and utilizes a fusion algorithm to perform time-series alignment and weighting of data from different sources to obtain real-time external environmental brightness and short-term predicted values. This multi-source fusion acquisition method significantly improves the accuracy and robustness of environmental lighting information, ensuring that the system can quickly and accurately capture changes in external lighting, providing a reliable data foundation for subsequent dimming control, and achieving a dynamic response of "predictive adjustment." This effectively avoids problems such as light output delay or overshoot caused by changes in the external environment, thereby maintaining a smooth and stable lighting environment at the tunnel entrance.

[0024] 2. Intelligent, Segmented Dimming Control: This invention divides the tunnel's internal lighting system into several independent dimming zones, deploying vehicle detectors and dimming controllers within each zone. This enables dynamic segmented dimming based on real-time vehicle status and traffic flow. Specifically, when a vehicle is detected about to enter a zone, a feedforward and closed-loop control algorithm increases the zone's brightness from a low-power standby state to a level sufficient for safe driving. After the vehicle leaves the zone, and after a delay confirming no vehicles remain, the zone automatically returns to standby mode. This control strategy enables tiered dimming for different areas (entrance, transition, basic, and exit), meeting safety lighting requirements while significantly reducing energy consumption and improving overall energy efficiency.

[0025] 3. Lamp Light Decay Compensation and Closed-Loop Adaptive Adjustment: LED lamps experience light decay due to aging and dust accumulation during long-term use, leading to a decrease in actual output brightness. To address this, this invention introduces a lamp light decay model and incorporates a light decay compensation factor into the dimming closed-loop control. This allows the dimming command calculated by the stepless dimming algorithm to automatically compensate for the performance degradation of the lamp. By comparing the error between the actual measured brightness and the target brightness, and using feedforward and PI control (or fuzzy control) methods for closed-loop correction, the theoretical output brightness is ensured to remain stable at a level meeting specifications over the long term, thereby improving the overall stability and reliability of the lighting system.

[0026] 4. Enhanced System Safety and Fault Tolerance: This invention incorporates abnormal state handling and redundancy mechanisms. When the vehicle detection or brightness detection sensor malfunctions, it can automatically revert to a preset stepless dimming or sequential dimming mode, ensuring sufficient safety lighting is maintained within the tunnel at all times. Simultaneously, in the event of emergencies such as fires or accidents, the system can immediately trigger an emergency lighting mode, setting the illuminance of critical areas to the highest level (100%), providing ample visual support for drivers and ensuring safe evacuation. This multi-layered safety protection mechanism significantly enhances the system's reliability and robustness.

[0027] 5. Energy Saving Effect and Adaptive Optimization: This invention collects power data for each dimming zone in real time and compares it with the energy consumption benchmark under full-power constant illumination using an energy consumption integral model to calculate the energy saving rate. Through data statistics and feedback, this system can adaptively optimize dimming parameters and iteratively upgrade the control strategy by combining digital twin and self-learning algorithms. Under the premise of ensuring driving safety, it significantly reduces tunnel lighting energy consumption, achieving significant energy saving effect, thereby reducing operating costs and extending equipment life.

[0028] 6. Intelligent Human-Machine Interaction and Remote Management: The system is equipped with a unified monitoring platform and remote management interface, allowing maintenance personnel to view real-time lighting data, traffic flow status, and energy consumption statistics for each area. Through the "one-click full illumination" or manual adjustment functions, rapid intervention in emergencies is possible. This human-machine interaction platform not only enhances the intelligence level of system management but also provides data support for subsequent maintenance and upgrades, further ensuring the safety and efficiency of tunnel lighting control.

[0029] In summary, this invention integrates multi-source environmental monitoring, vehicle detection, light decay compensation, and intelligent dimming technologies to achieve closed-loop control and adaptive energy saving of the tunnel lighting system from entrance to exit. Attached Figure Description

[0030] Figure 1 This is a system flowchart of the present invention.

[0031] Figure 2 This is a system flowchart for acquiring brightness outside the cave and adjusting the brightness at the entrance.

[0032] Figure 3 This is a system flowchart for vehicle detection and segmented control.

[0033] Figure 4 It provides closed-loop and stepless dimming for the brightness inside the cave.

[0034] Figure 5 This is a schematic diagram of the system structure used in the method of the present invention.

[0035] Figure 6This relates the daily energy-saving data collected by the method of this invention to traffic flow and time period.

[0036] Figure 7 This is the platform's data analysis interface. Detailed Implementation

[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present invention.

[0038] like Figure 1 The method for intelligent management and control of energy-saving lighting in highway tunnels includes the following steps: 1) external brightness acquisition and entrance dimming; 2) vehicle detection and segmented control; 3) internal brightness closed-loop and stepless dimming; 4) abnormal state handling and redundancy mechanism; 5) data aggregation and energy-saving statistics; 6) manual intervention and management. Each step is explained in detail below.

[0039] I. External brightness acquisition and entrance dimming The external ambient brightness data at the tunnel entrance is acquired using multi-source sensing devices, including brightness detectors, video image recognition, and optional weather and sunshine forecast information. The multi-source observations are then time-series aligned and fused to obtain the real-time external tunnel brightness. Lout (t) and short-term forecasts The fusion results are combined with tunnel lighting design specifications to obtain the target brightness of the entrance section. When detected When a rapid increase or decrease is imminent, the lighting power of the entrance section is adjusted in advance using a feedforward + closed-loop dimming algorithm to avoid the black hole effect or over-illumination.

[0040] 1. Observation and Forecasting The following observation sources are deployed outside the tunnel entrance: a dedicated brightness detector at a fixed angle is installed at a certain distance outside the entrance to periodically output measured values. ; The environment at the tunnel entrance is captured by a camera, and an image brightness recognition algorithm is used to obtain... ; Obtain the solar radiation forecast for the next 1-3 hours from the meteorological platform API. Or cloud cover rate, rainfall probability index; combine the above data at time t to form the original observation: .

[0041] 2. Multi-source fusion right After filtering and time alignment (if there are different sampling frequencies), a weighted method or neural network is used for fusion to output the fused out-of-hole brightness: , in For linear weighting, Kalman filtering, or small multilayer perceptron (MLP).

[0042] 3. Short-term forecast To anticipate changes in external brightness, a short-term prediction of Lout(t) is performed: , in f pred ARIMA, LSTM, or simplified models based on weather forecast rules are used; δ represents a number of historical time intervals; kΔ is the forecast interval, and Δ is the forecast step size (e.g., 30 seconds, 1 minute, etc.).

[0043] 4. Target brightness reference at the entrance According to the "Detailed Specifications for Lighting Design of Highway Tunnels", the reference brightness for the entrance section is... satisfy: , It also considers correction factors such as vehicle speed and visual adaptation coefficient. More complex calculations can be performed based on actual brightness outside the tunnel, traffic flow, and slope shading conditions. This method introduces the following predicted quantities: , Where Ф(.) is the prediction correction function, if If a significant increase is imminent, set α to a larger value at the current moment, and vice versa.

[0044] 5. Stepless dimming and predictive feedforward Set up a dimming signal for the lights at the tunnel entrance section. u entr (t)∈[0,1], corresponding to the lamp power ratio; in order to preemptively adjust when the brightness outside the cave is about to change, a feedforward + closed-loop dimming algorithm is adopted: 1) Feedforward term:

[0045] like If the light output indicates that the brightness will drop sharply in the next moment, the output of the entrance lights will be increased in advance; conversely, if the light output indicates that the sunlight will become stronger, the output will be reduced in advance.

[0046] 2) Closed-loop term: Simultaneously based on the actual brightness of the entrance segment or the expected brightness given by the management platform. Perform PI control: ,

[0047] make , The clip option restricts the result to the range of 0 and 1.

[0048] 3) Smooth transition and segmented linkage Smoothing function: To avoid frequent and rapid changes in entrance lighting fixtures, a smoothing function is used... u entr (t) Add speed limiter or small filter:

[0049] λ∈(0,1) controls the degree of smoothness.

[0050] When the stepless dimming variation in the entrance section is large, the transition section can be adjusted accordingly to avoid a "brightness cliff" between the opening and the transition section. This can be achieved by performing another interpolation or difference adjustment at the platform layer or edge layer.

[0051] This method proposes a comprehensive solution for "external tunnel brightness acquisition and entrance dimming," incorporating multi-source fusion, predictive feedforward, and closed-loop adaptive dimming. By combining mathematical models with practical engineering logic, it can: 1. Significantly improve the accuracy and response speed of external tunnel brightness detection; 2. Reduce lag or overshoot problems caused by rapid changes in ambient light; 3. Smooth the brightness transition between tunnel entrance and transition sections, improving driving comfort and safety; 4. Take into account differences in time periods with and without vehicles and changes in weather conditions, providing tunnel entrance brightness that meets specifications while reducing lighting energy consumption.

[0052] II. Vehicle Inspection and Segmented Dimming Control At the tunnel entrance, vehicle information is collected using AI cameras, millimeter-wave / LiDAR, or road studs; the tunnel interior lighting is divided into several independent dimming zones at preset intervals, and a vehicle detector and dimming controller are deployed in each dimming zone; when a vehicle is detected about to enter a dimming zone, that zone is switched from standby brightness in advance. B standby Adjust to vehicle brightness B car (z,t) After the vehicle leaves and it is confirmed that no new vehicle is waiting, the section will be restored to standby brightness. The core innovation of this solution lies in the integration of multiple sensors (including V2X vehicle-road cooperation), the introduction of short-term traffic flow prediction, and the adoption of distributed cooperative control and self-learning mechanisms, thereby achieving better comprehensive benefits in terms of safety and energy saving.

[0053] (I) General Description of the Method 1. Multi-source hyperfusion detection Multiple sensors are deployed at the tunnel entrance and inside the tunnel, including AI video cameras, millimeter-wave or LiDAR, geomagnetic sensors, and V2XRSUs (roadside units) that interact with the vehicle's OBU (On-Board Unit). Deep data fusion algorithms (such as those based on Transformer or bidirectional GRUs) are used to perform unified time alignment of information from different sensors and the vehicle's own data (license plate number, speed, estimated arrival time, etc.) to generate an initial record of the vehicle's status.

[0054] 2. Segmented dimming and distributed control The tunnel is divided into several dimming sections, each equipped with an integrated device of "vehicle detector + dimming controller" with local intelligent (micro neural network or rule-based self-learning) function; when a vehicle enters or is about to enter the section, the brightness is increased, and after a certain delay after leaving, if no new vehicle enters, the brightness is restored to standby mode. The lighting transition between sections is achieved through "neighborhood collaboration".

[0055] 3. Traffic flow prediction and adaptive parameter tuning The backend (or edge server) uses a time-series prediction model (such as LSTM / GRU) to predict short-term traffic flow based on historical traffic flow, weather, and holiday information. Then, it automatically adjusts the "advance light-on distance" and "delayed light-off duration" for each segment based on the prediction results. When the actual vehicle speed deviates from the design speed, the timing of the light-on triggering in the next segment can be dynamically changed to avoid insufficient or wasted lighting due to speeding or low speed.

[0056] 4. Self-learning and digital twins In the background, reinforcement learning or evolutionary algorithm analysis is performed on the tunnel operation data to optimize the control parameters of segmented dimming; a digital twin simulation environment is established to simulate and evaluate the new strategy; if the safety and energy-saving indicators are good, the strategy is then implemented in each section of the tunnel.

[0057] (II) Mathematical Formula Model For simplicity, a discrete-time model is used: the system updates its state and control every Δt seconds. Let t... k =k Δt,t k =0,1,2,…

[0058] 1. Vehicle trajectory and vehicle-road cooperative model 1.1 V2X Vehicle Information For vehicles equipped with roadside unit V2X functionality, the information packet they send before entering the tunnel is recorded as follows: , ID i For license plate or OBU identification; t entry This refers to the estimated time to enter the tunnel; v est、 aest These are the vehicle's estimated speed and acceleration, respectively, from its built-in sensors. At this point, the tunnel system can predict: ; This is to "light up" the relevant sections before the vehicles actually enter the tunnel.

[0059] 1.2 Multi-source observation fusion If the vehicle does not have a roadside unit (V2X), the initial observation Zi is obtained through AI video, radar, and geomagnetic sensors at the entrance, containing the license plate ID or temporary ID, driving speed, and vehicle type. At time tk, each sensor outputs a set of observations. Each observation can be represented as ; It may not include all fields, depending on the sensor's functionality.

[0060] In order to correlate these observations with the vehicle state {X} i (t k For association, use "minimum distance matching" or "neural network association": , in For the predicted position and velocity of vehicle i, Ω and Λ respectively measure the difference in position and velocity; if di,m≤γ, then observation m is regarded as information of vehicle i; unmatched observations are used to create new trajectories.

[0061] 1.3 State Update and Prediction For each vehicle i, define a state vector.

[0062] x i (t k This represents the cumulative distance the vehicle travels within the tunnel. v i (t k () represents the longitudinal velocity; a i (t k ) is acceleration (optional, used for more accurate trajectory prediction).

[0063] make This represents a prediction for the next moment (e.g., a uniform acceleration model):

[0064] , If there is no clear acceleration observation, it can be simplified to a uniform velocity model a. i =0.

[0065] When there are new observations Z m When associated with vehicle i, the state is corrected using a fusion method (such as Kalman or weighted average): .

[0066] 1.4 Segment Mapping and Neighborhood Collaboration The tunnel length is [0, L] total Divide the data into N segments, with each segment z having a range of [S]. z−1 , S z );make z i (t k If x = z i (t k )∈[Sz−1, Sz), To achieve a smooth transition at the edge of a segment, when the vehicle is less than a certain threshold D from the boundary of the next segment... offset At this time, a "vehicle is about to enter" or "warning" signal can be sent to the next section controller.

[0067] 2. Segmented dimming and adaptive strategy 2.1 Definition of Brightness ∈[0,1]: The brightness ratio of segment z at time tk; Standby brightness: ≈0.1~0.3B_, when there are no trains in the section; Vehicle brightness: Based on the brightness outside the cave Traffic flow prediction Section Standard Awaiting decision:

[0068] Where Φ is constructed by a neural network or interpolation function, ensuring that the result is in [0,1].

[0069] 2.2 Early lighting and delayed dimming Pre-lighting: When vehicle i predicts it will enter segment z in δt seconds, and δt ≤ toffset, the dimming controller for segment z is set...

[0070] Delayed dimming: If the number of vehicles detected in a section changes from non-zero to 0, a countdown will begin. If no vehicle enters within time τz, then

[0071] If the new car arrives before the countdown ends, the countdown will be interrupted, and the car's brightness will remain on.

[0072] 2.3 Distributed Cooperative Controller The dimming controller for each section is denoted as Periodic Δt execution: With adjacent sections ( , Exchange status, such as vehicle queue information for this section, current brightness value, and prediction for the next time moment; Calculate the brightness of this section based on vehicle location, external tunnel brightness, and traffic flow prediction; If a vehicle is detected to be speeding (v) i vd esign If +∈, then t will automatically increase. offset Or D offset ; Even when network latency occurs or commands are not received, each segment can still independently decide whether to "light up" or "dark up," ensuring system resilience.

[0073] 2.4 Self-learning and traffic flow prediction Short-term traffic flow prediction: On a backend or edge server, historical traffic flow and weather / holiday features of {Q(t-Δ),…,Q(t-1)} are used as input to LSTM / GRU to obtain the forecast. .

[0074] Self-learning parameter update: The system sets a control parameter vector. The vector can be iteratively optimized using reinforcement learning (RL) or evolutionary algorithm (GA) to improve energy efficiency or security performance.

[0075] III. Closed-loop brightness control and stepless dimming inside the cave This method deploys high-precision brightness detectors and self-learning dimming controllers, including IoT dimming controllers and radar-type dimming controllers, in multiple key areas within the tunnel (entrance reinforcement section, transition section, intermediate section, exit section, etc.), forming an intelligent brightness control system that integrates "real-time closed-loop measurement + stepless dimming + predictive correction".

[0076] (I) The overall approach and key innovations are as follows: 1. Multi-point brightness detection High-precision brightness detectors are deployed at at least four key sections within the tunnel: entrance E, transition T, main tunnel B, and exit X, to collect real-time "actual brightness" data. Each detector should be calibrated and periodically checked in advance to ensure low relative measurement error and resistance to interference from light fixtures or wall reflections.

[0077] 2. Considering luminous decay and dynamic maintenance coefficient of lamps An online evaluation of tunnel luminaires is conducted using a light decay model to estimate the actual light output reduction caused by time or degree of dirt accumulation. This model is used to correct the discrepancy between the "target brightness" and the "actual brightness," ensuring that the system can still accurately perform closed-loop dimming even after the luminaires have aged or accumulated dust.

[0078] 3. Integrate external brightness with standard requirements The lighting levels for entrance and transition sections often depend on the ambient brightness: entrance lighting should be increased during the day when ambient brightness is high, and appropriately reduced at night or in overcast weather. Reference illuminance for each section is dynamically adjusted based on external brightness data collection and relevant tunnel design specifications (JTG / TD70 / 2-01, etc.).

[0079] 4. Adaptive stepless dimming algorithm Discrete-time (period Δt) closed-loop control is employed, and each key segment is divided into "predictive feedforward + closed-loop PI (or fuzzy control) + light decay correction" to output stepless dimming levels ∈ [0,1] or divided into higher discrete levels (such as 100 levels). This ensures that the minimum illuminance required by the specifications is met while avoiding "over-illumination".

[0080] (II) Mathematical Model The following is the core formula for tunnel brightness closed-loop and stepless dimming. High-precision brightness detectors are deployed at at least four key sections within the tunnel: entrance E, transition T, basic B, and exit X, to collect real-time actual brightness data. .

[0081] 1. The brightness state and light decay model are as follows: 1) Actual brightness measurement remember Let time t k The pavement brightness value (cd / m²) measured in section z for k=0,1,2,… 2 ).

[0082] 2) Light decay coefficient remember The light decay ratio of the luminaire in segment z is estimated from a prior light decay model or detection data: , F is an empirical formula or a function of cumulative running time, temperature, and dust deposition factors.

[0083] 3) Relationship between nominal output brightness and actual brightness When the controller issues a relative brightness command When (i.e., stepless dimming power ratio), the nominal output brightness is set to (Can be a linear or nonlinear function), considering light decay, the theoretical brightness reaching the road surface is: .

[0084] 2. Reference brightness and external environment 1) Brightness outside the cave remember This represents the real-time measured brightness outside the tunnel; 2) Standardized segmentation requirements According to the "Detailed Rules for Lighting Design of Highway Tunnels", a minimum target brightness is defined for each section z. Common situations: Entry section: ; Transition section: Decreasing function +Standardized fixed coefficients; Middle section: ≥ It can be finely adjusted according to traffic flow; Export section: ≥ The influence of external brightness is also taken into account.

[0085] 3) Overall reference brightness , Where L safe It is a safety baseline value to ensure that the minimum video surveillance and emergency response requirements are met.

[0086] 3. Stepless dimming control model make e z (t k ) Indicates the brightness error of segment z: , in, That is, the actual measured brightness.

[0087] To ensure fast and stable dimming, a combination of feedforward and closed-loop methods is used: 1) Feedforward term: Based on the estimated light decay coefficient and nominal brightness curve, first calculate the following: , Where Γ = is a calibration constant relating the rated maximum brightness of the luminaire to the dimming signal; 2) Closed-loop correction term: Use a PI converter to adjust the error. e z (t k ) Make corrections; PI method: ; Where K p ,K i For proportional or integral gain, adaptive or online scheduling; 3) Output limiting:

[0088] To avoid frequent fluctuations in dimming commands, perform index smoothing or speed limiting operations: ; Where λ∈(0,1) controls the smoothness, and clip restricts the instructions to the range [0,1].

[0089] IV. Abnormal State Handling and Redundancy Mechanism When the vehicle detection sensor malfunctions or goes offline, it automatically reverts to stepless dimming or sequential dimming mode to maintain basic lighting; when the brightness detector malfunctions or goes offline, it automatically switches to the preset sequential dimming mode; when a fire or other emergency event is detected, the emergency lighting mode is triggered, switching the relevant section or the entire tunnel to the highest brightness.

[0090] 1. Abnormal Status Monitoring Module 1.1 Sensor Status Monitoring To achieve automatic fault tolerance, the system includes a sensor status monitoring module, whose main function is to periodically monitor the operating status of each sensor. For each vehicle detection sensor and brightness detector, the following state variables are defined: For vehicle detection sensors: .

[0091] For brightness detectors:

[0092] In practical implementation, a "data update time threshold" T can be set. th (5 seconds or 10 seconds), if T continues th If there is no valid data, the sensor is considered to be in an abnormal state, and F(t) = 1.

[0093] 1.2 Emergency Status Detection Simultaneously, the system is equipped with an emergency status monitoring module to receive signals from fire alarms, accident detection (through video analysis, smoke detection, temperature rise, and other emergencies). Emergency flags are set:

[0094] These status data are aggregated in real time by local controllers and the central management platform to trigger corresponding control strategies.

[0095] 2. Exception handling logic In the software logic, based on the sensor status and emergency indicators, the system formulates the following decision rules: 2.1 Troubleshooting Vehicle Detection Sensors Detection logic: For a certain dimming section, if the corresponding vehicle detection sensor F... veh (t) = 1 (that is, at the specified time T) th If no valid vehicle data is received, a fault handling action will be triggered.

[0096] Processing strategy: Automatically switch to "stepless dimming" or "sequential dimming" mode.

[0097] Stepless dimming mode maintains the preset basic brightness Bs tandby Alternatively, maintain the recent brightness level based on historical data; The time-based dimming mode sets a fixed light output according to a preset time schedule (based on day-night switching and seasonal changes).

[0098] Define the dimming mode signal M(t) as:

[0099] In M fallback Under this condition, the corresponding output brightness B(t) of the lamp is: , Among them B seq (t) represents the preset timing dimming value.

[0100] 2.2 Brightness Detector Troubleshooting Detection logic: When the brightness detector F of a certain key lighting node... lum When (t)=1, it means that the detector is currently unable to provide actual brightness data.

[0101] Processing strategy: Automatically switch to the preset timing dimming mode and maintain illumination by using pre-stored historical brightness or control parameters based on astronomical clocks and preset curves.

[0102] In mathematics, we can let: B(t) = B time (t), Among them B time (t) is determined by a preset timing table, such as "full brightness during the day, low brightness at night".

[0103] 2.3 Emergency Response Detection logic: When the emergency status monitoring module detects a fire or other major accident, i.e., E... mergency When (t)=1, the system immediately enters an emergency state.

[0104] Handling strategy: For the affected section or the entire tunnel, immediately trigger emergency lighting, adjusting the lights in the corresponding area to maximum brightness (usually set to 100% output) to ensure sufficient illumination at the accident site for rescue and personnel evacuation. Mathematically defined: B(t) = 1.0, ∀affected area, and after the accident is resolved, gradually return to the original automatic dimming state through manual command or automatic recovery.

[0105] V. Data Summary and Energy Saving Statistics The system uploads time-series data collected from vehicle detection, brightness detection, and dimming controller to the backend; compares the energy consumption of the constant-on mode, calculates the energy-saving rate, and outputs an energy-saving report; and adaptively optimizes the dimming parameters based on the statistical results, combining digital twin or self-learning algorithms to improve overall lighting efficiency.

[0106] (a) Data collection and uploading 1. Data collection for each module Vehicle detection module: Each detection device (AI camera, millimeter-wave radar, geomagnetic sensor) collects vehicle passing data (license plate number, vehicle speed, number of vehicles, location, etc.) at a fixed interval (sampling once per second); Brightness detection module: A set of high-precision brightness detectors is installed in key areas such as the tunnel entrance, transition, main tunnel, and exit to collect real-time actual brightness data of the area. z meas (t), where z represents different control sections of the tunnel; Dimming control module: The dimming controllers in each section record the output dimming commands B in real time. z (t), which is between 0 and 1, represents the ratio of the actual output power of the lamp to its full power.

[0107] 2. Data Upload and Synchronization The time-series data from each module are uploaded to the back-end management system through a reliable communication network (wired Ethernet, wireless LoRa, 4G / NB-IoT, etc.). The back-end system timestamps the data, stores it in a formatted manner, and categorizes it according to the region ID to form a unified time-series database.

[0108] (II) Data Statistics and Energy Consumption Report Generation 1. Data Summary The backend management system periodically extracts time-series data uploaded from each segment of the database, including vehicle detection, brightness detection, and dimming control data. Through data preprocessing and cleaning, noise and abnormal data are filtered out or corrected to ensure data accuracy.

[0109] 2. Energy consumption statistical analysis Based on the aforementioned energy consumption integral model, the actual energy consumption and theoretical constant-on energy consumption of each section and the entire tunnel are calculated. Simultaneously, the energy-saving rate is statistically calculated, and an energy consumption statistical report is generated over a certain period (daily, weekly, or monthly). This report includes: real-time power curves and cumulative energy consumption data for each section; a comparative analysis of the energy consumption differences between the current dimming strategy and the constant-on mode; an energy-saving rate trend chart; and alarm prompts (if the energy-saving rate is below a preset threshold).

[0110] (III) Adaptive optimization and digital twin support 1. Adaptive optimization mechanism Based on the energy consumption report and historical statistical results of data from various sensors, the backend uses adaptive optimization algorithms (reinforcement learning, evolutionary algorithms, or neural network models) to optimize the dimming parameters.

[0111] Define the control parameter vector Θ={K p ,K i ,κ,t delay ,t offset ,…}; The optimization objective is to maximize the energy saving rate η while ensuring that the actual illuminance in each section remains stable within the target range.

[0112] Through real-time feedback and digital twin simulation systems, existing dimming strategies are virtually simulated and optimized offline to form new optimal control parameter combinations, which are then periodically distributed to each dimming module.

[0113] 2. Digital twin technology A digital twin model corresponding to the actual tunnel lighting system is constructed, simulating the system's energy consumption, light decay, and traffic flow dynamics. The digital twin platform is used to simulate and verify different dimming strategies, predicting the system's performance under various environmental conditions and providing a basis for adaptive optimization decisions.

[0114] VI. Manual Intervention and Management The monitoring platform provides manual dimming and "one-click full brightness" functions to deal with sudden weather or emergencies; when maintenance personnel intervene, any section can be temporarily switched to the highest brightness or restored to automatic control.

[0115] 1. System Architecture The manual intervention and management functions mainly consist of the following parts: 1) Monitoring platform interface The human-machine interface (HMI) displays real-time brightness, traffic flow status, and energy consumption data for each section to operators via PC, tablet, or other smart terminals, and provides operation buttons such as "one-click full brightness", "manual dimming", and "restore automatic mode".

[0116] 2) Command sending module When operators issue manual intervention commands on the monitoring platform, the module sends the commands to the dimming controllers in each section via secure and reliable communication protocols (such as TCP / IP, MQTT, etc.).

[0117] 3) Dimming controller After receiving a manual intervention command, the dimming controllers deployed in each section will prioritize the use of the manual mode parameters specified by the monitoring platform (the full brightness command is dimming ratio B=1.0) to cover the automatic dimming strategy. At the same time, the dimming controllers support the storage of preset automatic dimming parameters. When the operator executes the "restore automatic" command, it will switch back to the original automatic closed-loop dimming mode.

[0118] 4) After receiving a manual command, the dimming controller of the status feedback module will feed back the current working status and actual brightness data to the monitoring platform to ensure that operators can keep abreast of the system status in real time and make necessary adjustments or confirmation operations.

[0119] like Figure 5 As shown, the Jianjin Mukeli Tunnel was selected as the pilot application site for this invention project. The left tunnel is 2575m long, and the right tunnel is 2594m long. The Mukeli Tunnel on the Jianjin section of the Changchun-Shenzhen Expressway (G25) was used as the pilot tunnel for investigation. LED tunnel lights were used, with 50W lights in the middle section and 200W, 80W, and 50W lights in the reinforced sections, all symmetrically arranged on both sides at an installation height of 5.4m. After the system debugging of this invention was completed on January 26, 2024, the on-board dimming control strategy was prioritized in both the left and right tunnels of the Mukeli Tunnel. During this period, if the brightness detector went offline, the system reverted to sequential dimming; if the vehicle detection sensor went offline, it reverted to stepless dimming. For nearly two months, different dimming modes were run in the left and right tunnels for comparative energy saving. The following sections describe the energy-saving effect of on-board dimming from two aspects: energy saving comparison of different modes in the left and right tunnels and energy saving comparison with historical electricity costs.

[0120] 1. Energy-saving comparison of different modes for left and right tunnels Investigations revealed that the average daily traffic volume in the left and right tunnels of the Mukoli Tunnel was consistently similar. Therefore, a comparative energy consumption method between the left and right tunnels was used to compare the energy-saving effects of the on-board dimming control mode. To avoid the impact of uneven traffic flow distribution and weather conditions outside the tunnel on lighting energy consumption, this comparison period could be appropriately extended. The test period was selected as 73 days, from July 8th to September 18th. The dimming strategies for the two tunnels are as follows.

[0121] Upbound (Hangzhou -> Jinhua) Referring to the dimming strategies commonly used in Jinhua, Zhejiang, such as the tunnel lighting control scheme of the Mazhai Tunnel Management Office, the following upward dimming strategy is formulated based on the sunny day mode in summer (June-September): Increased lighting hours: 5:00 AM – 7:00 PM Enhance the brightness adjustment ratio of the lighting fixtures: Automatic (automatically correlated with the brightness outside the tunnel).

[0122] Basic lighting hours: All day Basic lighting brightness dimming ratio: Daytime: 90% brightness; Nighttime: 80% brightness.

[0123] Downstream (Jinhua -> Hangzhou) Using the vehicle-mounted dimming mode, the brightness levels of each lighting circuit and lighting section are dynamically adjusted based on the external brightness and real-time traffic flow.

[0124] 1) Strengthen the lighting activation strategy: The brightness outside the cave is higher than 90 cd / m 2 At that time, all enhanced lighting fixtures are turned on, and the brightness adjustment level is automatically adjusted according to the brightness outside the tunnel (meeting the requirements in the lighting design specifications). The brightness outside the cave is less than 90 cd / m 2 At that time, turn off all enhanced lighting fixtures.

[0125] 2) Basic lighting switch strategy: The external brightness of the cave is less than 38 cd / m² 2 When the traffic flow is less than 700veh / h, turn off the right-side basic lighting and keep the left-side basic lighting circuit on.

[0126] The external brightness of the cave is higher than 38 cd / m 2 Or when the traffic flow is greater than 700veh / h, turn on the right-side basic lighting.

[0127] 3) Brightness dimming ratio strategy: When there are vehicles present, the brightness within approximately 1.5 times the lighting visibility distance in front of any vehicle meets the requirements of the D702-01 Highway Tunnel Lighting Design Specifications. The brightness adjustment level is adjusted based on actual on-site calibration data. When there are no vehicles present, the brightness adjustment level meets the requirements for video surveillance (15%).

[0128] 2. Test Results The following comparative data were obtained by calculating the meter data of each lighting circuit in the two holes, which were read in real time during the test.

[0129] Table 1 Energy Consumption Statistics

[0130] Note: The above data comes from real-time readings of 12,024 energy consumption meters.

[0131] According to the statistics in the table above, the energy saving ratio of the Jinhang direction during the comparison test of the left and right tunnels was 46.69%, with an average daily energy saving of about 214 kWh.

[0132] 3. Energy saving comparison with historical data The energy consumption data for the two-way lighting of the Keli Tunnel for 24 months from 2022 to 2023, which were statistically archived, are shown in the table below.

[0133] Table 2 Historical Energy Consumption Data

[0134] Calculations showed that the historical average daily lighting energy consumption for both tunnels was 1248.34 kW*h. The historical average daily lighting energy consumption for a single tunnel was 714.17 kW*h, which was used as the benchmark for energy saving comparison.

[0135] After project implementation in early 2024 and a period of system debugging, the system was put into operation on January 26, 2024. During this period, from July 8 to September 18, a two-month energy-saving comparison was conducted between different dimming modes in the left and right tunnels. The Hangzhou-Jinhua direction operated in sequential circuit dimming mode, while the Jinhua-Hangzhou direction operated in vehicle-mounted dimming control mode. During this period, the overall tunnel lighting energy consumption increased; at other times, both tunnels operated in vehicle-mounted dimming control mode. Statistics show that the system has been running stably for 306 days to date (November 26, 2024). Traffic flow data and energy consumption data for both tunnels are as follows: Table 3. Traffic Flow and Energy Consumption Statistics During Operation

[0136] Note 1: Dimming mode of sequence circuit during operation in the Hangzhou-Jinhua direction from 7 / 8 to 7 / 31.

[0137] Note 2: All the above data is exported from system logs.

[0138] Calculations show that the average daily lighting energy consumption for the Hangzhou-Jinhua route is 310 kWh, for the Jinhua-Hangzhou route it is 255 kWh, and for both directions it is 565.4 kWh. Compared to historical lighting energy consumption, the tunnel saves an average of 862.84 kWh per day (1248.34 - 565.4) in both directions, representing an energy-saving efficiency of approximately 60.4%. The annual energy saving is approximately 315,000 kWh. (The energy-saving ratio will be slightly higher after deducting data from the Hangzhou-Jinhua route's sequential dimming strategy during the two-month comparison period.)

[0139] 4. Energy saving analysis at different times of the day During the energy consumption comparison period around July 2024, an analysis was conducted on the relationship between energy saving at different times of the day and traffic flow and time period, as follows: Figure 6 As shown.

[0140] Following the numbered order in the above diagram, the analysis is as follows: 1) During the night, half of the basic lighting on the vehicle's lighting side is turned off, and the brightness is adjusted to just meet the monitoring needs. Compared with the test where the lights are not turned off but the overall brightness is reduced, the energy saving effect is about 75%. 2) During the dawn phase, the vehicle-mounted lighting is dynamically linked to the external brightness of the tunnel to enhance the lighting (the daytime and nighttime thresholds are calculated in detail and vary every day). A comparative test was conducted to trigger the enhanced lighting based on time. Because the vehicle-mounted lighting is enhanced for a period of time at night, the energy-saving effect reaches 85% for one hour. 3) During the daytime, the energy-saving effect of vehicle lighting is inversely proportional to the brightness outside the tunnel and the traffic flow. It is a dynamic change, while the comparison side is a fixed value. Therefore, the energy-saving ratio is also dynamic, with an energy-saving range of 45%-65%. 4) At critical nighttime, the vehicle-mounted lighting is dynamically correlated with the external brightness of the tunnel, and the enhanced lighting is turned off (the critical points for daytime and nighttime are calculated in detail and vary every day). The comparison test is conducted by triggering the enhanced lighting to turn off at a certain time. Since the enhanced lighting is turned off for a period of time in the morning, the energy-saving effect reaches 80% for one hour. 5) Late at night, same as 1; 6) Dawn stage, same as 2.

[0141] 5. Energy consumption viewing and recording On the platform's energy-saving analysis interface, select the comparison time period and the corresponding sidebar to display detailed bidirectional energy consumption comparison data within that time period. Data also includes real-time traffic flow, real-time tunnel exterior brightness, and real-time energy consumption curves for reference. (See below.) Figure 7 As shown, the data export function can be used to export all energy consumption, traffic flow, and other data. The traffic flow data, tunnel brightness data, and energy consumption data mentioned above can all be viewed and exported on this interface.

Claims

1. A method for intelligent control of energy-saving lighting in highway tunnels, characterized in that, Includes the following steps: 1) External brightness acquisition and entrance dimming: The external ambient brightness data of the tunnel entrance is acquired through multi-source sensing devices, including brightness detectors, video image recognition, and optional weather and sunshine forecast information; The multi-source observations are time-series aligned and fused to obtain the real-time external brightness of the hole. L out (t) and short-term forecasts ; The fusion results were combined with tunnel lighting design specifications to obtain the target brightness of the entrance section. When detected When a rapid increase or decrease is imminent, the lighting power of the entrance section is adjusted in advance using a feedforward + closed-loop dimming algorithm to avoid the black hole effect or over-illumination. 2) Vehicle inspection and segmented control: Vehicle information is collected at the tunnel entrance, and the tunnel's internal lighting is divided into several independent dimming zones at preset intervals. A vehicle detector and dimming controller are deployed in each dimming zone. When a vehicle is detected about to enter a dimming zone, that zone is switched from standby brightness in advance. B standby Adjust to vehicle brightness B car (z,t) After the vehicle leaves and it is confirmed that no new vehicle is left behind, the brightness of the section will be restored to standby mode. 3) Closed-loop brightness control and stepless dimming within the cave: Brightness detectors were installed at the entrance, transition, middle, and exit sections to measure the actual brightness of each section in real time. ; The dimming coefficient was corrected based on the luminous decay model of the lamps, and combined with the brightness outside the cave. L out (t) or traffic flow information, using a stepless dimming algorithm to perform feedforward + closed-loop control on the output power of each segment, so that the actual brightness meets the target brightness required by the specification. ; 4) Abnormal state handling and redundancy mechanisms: When the vehicle detection sensor malfunctions or goes offline, it automatically reverts to stepless dimming or sequential dimming mode to maintain basic lighting; when the brightness detector malfunctions or goes offline, it automatically switches to the preset sequential dimming mode. When a fire or other emergency is detected, the emergency lighting mode is triggered, switching the relevant section or the entire tunnel to its highest brightness.

2. The method according to claim 1, characterized in that, The multi-source fusion and prediction in step 1) includes: 1.1) Deploying the following observation sources outside the tunnel entrance: A dedicated brightness detector at a fixed angle is installed at a certain distance outside the entrance, and it periodically outputs the measured values. ; The environment at the tunnel entrance is captured by a camera, and an image brightness recognition algorithm is used to obtain... ; Obtain the solar radiation forecast for the next 1-3 hours from the meteorological platform API. Or cloud cover rate, rainfall probability index; The above data are combined at time t to form the original observation: ; 1.2) Through the fusion function f fusion Calculate the real-time external brightness L out (t); 1.3) Employing a time-series forecasting model f pred For L out (t) is used to make short-term predictions and obtain This is used to perform feedforward dimming on the entrance section lights; And / or, the inlet dimming in step 1) adopts the following mathematical model: Set entrance section dimming command u entr (t)∈[0,1]; Feedforward term: ； Closed-loop term: ， ； Dimming command limit: ， And mutations can be suppressed through a smoothing function.

3. The method according to claim 1, characterized in that, Step 2) includes the following steps: 2.1) Multi-source hyperfusion detection Multiple sensors are deployed at the tunnel entrance and inside the tunnel, including: AI video cameras, millimeter-wave or lidar, geomagnetic sensors, and roadside units that interact with the vehicle's onboard unit; deep data fusion algorithms are used to perform unified time alignment of information from different sensors and the vehicle's own information to generate an initial record of the vehicle's status; 2.2) Segmented dimming and distributed control The tunnel is divided into several dimming sections, and each section is equipped with a vehicle detector and dimming controller integrated device with local intelligent functions. When a vehicle enters or is about to enter the section, the brightness is increased. After leaving, if there is no new vehicle after a certain delay, the brightness is restored to standby mode. The lighting transition between sections is achieved through neighborhood collaboration. 2.3) Traffic Flow Prediction and Adaptive Parameter Tuning The backend or edge server uses a time-series prediction model to predict short-term traffic flow based on historical traffic flow, weather, and holiday information. Then, it automatically adjusts the advance lighting distance and delay of the dimming time for each section based on the prediction results. When the actual speed of the vehicle deviates from the design speed, it dynamically changes the lighting triggering time of the next section to avoid insufficient or wasted lighting due to speeding or low speed. 2.4) Self-learning and digital twins In the background, reinforcement learning or evolutionary algorithm analysis is performed on the tunnel operation data to optimize the control parameters of segmented dimming; a digital twin simulation environment is established to simulate and evaluate the new strategy; if the safety and energy-saving indicators are good, the strategy is then implemented in each section of the tunnel.

4. The method according to claim 3, characterized in that, In step 2): a) Set the tunnel length [0, L] total The area is divided into N dimming zones, each equipped with a vehicle detector and a dimming controller. b) When a vehicle approaches the section boundary by a certain advance distance, the system will switch the corresponding section from standby brightness to vehicle-accessible brightness. c) After the vehicle leaves, if the section is within the delay time t delay If no new car enters, the brightness will revert to standby mode (B). standby ; d) Adjacent sections can exchange detection data to prevent excessive brightness jumps; And / or, in step 2): a) For vehicles equipped with a roadside unit (OBU), obtain [data / information] before they enter the tunnel. ; b)ID i For license plates or OBU identification; t entry This refers to the estimated time to enter the tunnel; v est、 a est These are the vehicle's estimated speed and acceleration, respectively, from its built-in sensors. c) If the vehicle does not have a roadside unit, initial observations Z are obtained through AI video at the entrance, radar, and geomagnetic sensors. i At time t k Data association is performed internally; d) Predict using uniform acceleration or uniform velocity models And update the vehicle state X with minimum distance matching or neural network correlation when new data is observed. i (t k ); e) Segment mapping rule: If x i (t)∈[S z−1 If the vehicle is in section z and the dimming controller of section z is responsible for the dynamic dimming command, then the vehicle is in section z.

5. The method according to claim 1, characterized in that, Step 3) Deploy high-precision brightness detectors at at least four key sections within the tunnel: entrance E, transition T, base B, and exit X, to collect real-time actual brightness data. ; The steps include the following: 3.1) The brightness state and light decay model are as follows: remember Let time t k The road surface brightness values ​​measured in section z, k=0,1,2,… remember The light decay ratio of the luminaire in segment z is estimated from a prior light decay model or detection data: ， F is an empirical formula or a function of cumulative running time, temperature, and dust deposition factors; When the controller issues a relative brightness command At that time, the nominal output brightness is set to After considering light decay, the theoretical brightness reaching the road surface is: ； 3.2) Reference brightness and external environment remember This represents the real-time measured brightness outside the tunnel; according to the "Detailed Rules for Lighting Design of Highway Tunnels," a minimum target brightness is defined for each section z. Common situations: Entrance step: ; Transition section: Decreasing function +Standardized fixed coefficients; Middle section: ≥ It can be finely adjusted according to traffic flow; Export section: ≥ And take into account the influence of external brightness; Overall reference brightness: , Where L safe It is a safety baseline value to ensure that the minimum video surveillance and emergency response requirements are met; And / or, the stepless dimming control model in step 3) is as follows: make e z (t k ) Indicates the brightness error of segment z: , in, That is, the actual measured brightness; To ensure fast and stable dimming, a combination of feedforward and closed-loop methods is used: Feedforward term: Based on the estimated light decay coefficient and nominal brightness curve, first calculate the following: ， Where Γ is a calibration constant relating the rated maximum brightness of the luminaire to the dimming signal; Closed-loop correction term: using a PI to adjust the error e z (t k ) Make corrections; PI method: ； Where K p ,K i For proportional or integral gain, adaptive or online scheduling; Output limiting: ； To avoid frequent fluctuations in dimming commands, perform index smoothing or speed limiting operations: 。 6. The method according to claim 1, characterized in that, In step 4): When the vehicle detector fails to output data for ΔN consecutive times or the radar goes offline, it automatically reverts to stepless dimming mode and maintains the safe lower limit of tunnel illumination. When the brightness detector malfunctions, the system switches to sequential dimming or fixed power distribution scheduled by seasonal time periods. If a fire, accident, or emergency triggers E mergency When =true, max{B} is executed on the relevant section or the entire tunnel. z (t)}=1.0 is the highest brightness command.

7. The method according to claim 1, characterized in that, The method also includes uploading time-series data collected by vehicle detection, brightness detection, and dimming controller to the backend; comparing the energy consumption of the constant-on mode, calculating the energy-saving rate and outputting an energy-saving report; adaptively optimizing the dimming parameters based on the statistical results, and improving the overall lighting efficiency by combining digital twin or self-learning algorithms. And / or, the method also includes manual intervention and management: providing manual dimming and one-click full brightness functions on the monitoring platform to deal with sudden weather or emergencies; when maintenance personnel intervene, any segment can be temporarily switched to the highest brightness or restored to automatic control.

8. An intelligent control system for energy-saving lighting in highway tunnels, used to implement the method as described in any one of claims 1-7, characterized in that, include: a) External brightness acquisition module: A multi-source sensing device deployed outside the tunnel entrance to collect and fuse data for outputting L... out (t) and ; b) Vehicle detection and segmented control module, including sensors and dimming controllers deployed at the tunnel entrance and in each independent dimming section, used to output segmented lighting / darkening commands based on vehicle trajectory information; c) The tunnel brightness closed-loop and stepless dimming module is connected to the brightness detector of the key section of the tunnel. It is used to perform light attenuation compensation, feedforward + closed-loop dimming algorithm, and send dimming commands to the corresponding lamps. d) Abnormal state handling and redundancy mechanism module: When the detector fails or an emergency event is triggered, it automatically performs timed dimming or maximum brightness. e) Data aggregation and energy-saving statistics module, used to aggregate vehicle and brightness data, generate energy-saving reports, and adaptively iterate dimming strategies in conjunction with big data analysis; f) Manual intervention and management interface, providing human-computer interaction functions for the monitoring platform, allowing maintenance personnel to manually turn on all lights, make temporary corrections, or switch back to automatic control.

9. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the method of any one of claims 1-7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the method of any one of claims 1-7.

Images