Intelligent control method and system for tunnel lighting based on vehicle flow monitoring

Abstract

The embodiments of the present application relate to the field of intelligent control technology, and more specifically, to a method and system for intelligent control of tunnel lighting based on vehicle flow monitoring. The embodiments of the present application determine the lighting control strategy by obtaining various information of the target tunnel and performing a series of processing, and comprehensively consider three-dimensional space information, vehicle flow sensor monitoring information, and equipment network status information, so that the determination of the lighting control strategy is more comprehensive. The correlation is determined by using the debugged network to improve the accuracy of strategy matching. Identifying different vectors and performing fusion processing can carefully consider the mutual influence of various factors in the tunnel and ensure that the lighting control strategy adapts to the actual conditions of the tunnel. This method can determine the target lighting control strategy according to different correlations, which helps to improve the intelligence of the lighting system, thereby achieving energy saving, ensuring driving safety, and extending the service life of lamps, and improving the overall efficiency of tunnel lighting management.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of intelligent control technology, and more specifically, to a method and system for intelligent control of tunnel lighting based on vehicle flow monitoring. Background Art

[0002] There are many technical problems in traditional tunnel lighting control. In the past, lighting control often adopted a relatively single mode, such as adjusting the lighting brightness only according to a fixed schedule or a simple traffic flow threshold, which lacked comprehensiveness and accuracy. Since the three-dimensional spatial information of the tunnel is not fully considered, different sections (such as the entrance section, the middle section, and the exit section) may not be able to get appropriate lighting according to their special structure. At the same time, if the monitoring of vehicle flow is not deeply analyzed (the correlation is not determined through the debugged network as in the present invention), it is difficult to accurately match the lighting control strategy, which can easily cause excessive or insufficient lighting. Moreover, the equipment network status information is ignored. Once there is a problem with the equipment, the lighting control may be out of control. In addition, it is difficult for traditional methods to balance the needs of energy saving, driving safety, and lamp life. Summary of the invention

[0003] In order to at least overcome the above-mentioned deficiencies in the prior art, one of the objectives of the embodiments of the present application is to provide a method and system for intelligent control of tunnel lighting based on vehicle flow monitoring.

[0004] An embodiment of the present application provides a method for intelligent control of tunnel lighting based on vehicle flow monitoring, and the method is applied to an intelligent control system for tunnel lighting, and the method includes: obtaining three-dimensional spatial information of a target tunnel to be matched with a lighting control strategy, vehicle flow sensor monitoring information and device network status information of a vehicle flow monitoring device network corresponding to the target tunnel; wherein the vehicle flow sensor monitoring information is spatiotemporal domain monitoring information; the device network status information and the three-dimensional spatial information are non-spatiotemporal domain monitoring information; based on the vehicle flow sensor monitoring information, determining a first correlation between the target tunnel and a plurality of initial lighting control strategies respectively through a debugged first control strategy generation network; identifying at least one tunnel section space vector of the three-dimensional spatial information, and identifying at least one device network state vector of the device network state information; and generating a first control strategy after debugging. A second control strategy generation network is formed after the fusion of each lighting control decision vector, and a lighting intelligent control strategy vector of each lighting control decision vector is determined, and a lighting intelligent control strategy vector of each lighting control decision vector is determined respectively; according to the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy vector of each lighting control decision vector, each lighting control decision vector is fully connected to determine a second correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively; wherein the lighting control decision vector includes the at least one tunnel section space vector and the at least one device network state vector; and the target lighting control strategy of the target tunnel is determined by using the first correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively and the second correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively.

[0005] An embodiment of the present application also provides a tunnel lighting intelligent control system, including a processor and a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the above-mentioned tunnel lighting intelligent control method based on vehicle flow monitoring.

[0006] The embodiment of the present application also provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the above-mentioned intelligent control method of tunnel lighting based on vehicle flow monitoring is implemented. Beneficial Effects

[0007] The embodiment of the present application determines the lighting control strategy by obtaining various information of the target tunnel and performing a series of processing, comprehensively considering three-dimensional spatial information, vehicle flow sensor monitoring information and equipment network status information, so that the determination of the lighting control strategy is more comprehensive. The correlation is determined by using the debugged network to improve the accuracy of strategy matching. Identifying different vectors and performing fusion processing can carefully consider the mutual influence of various factors in the tunnel to ensure that the lighting control strategy adapts to the actual conditions of the tunnel. This method can determine the target lighting control strategy according to different correlations, which helps to improve the intelligence of the lighting system, thereby achieving energy saving, ensuring driving safety and extending the service life of lamps, and improving the overall efficiency of tunnel lighting management. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A flow chart of a method for intelligent control of tunnel lighting based on vehicle flow monitoring provided in an embodiment of the present application.

[0010] Figure 2 A block diagram of an intelligent tunnel lighting control system provided in an embodiment of the present application.

[0011] icon:

[0012] 100-Tunnel lighting intelligent control system;

[0013] 101 - processor; 102 - memory; 103 - bus. DETAILED DESCRIPTION

[0014] The exemplary embodiments disclosed in the embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the embodiments of the present application are shown in the accompanying drawings, it should be understood that the embodiments of the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the embodiments of the present application, and to enable the scope of the embodiments of the present application to be fully communicated to those skilled in the art.

[0015] In order to better understand the above technical scheme, the technical scheme of the embodiment of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiment of the present application and the specific features in the embodiment are detailed descriptions of the technical scheme of the embodiment of the present application, rather than limitations on the technical scheme of the embodiment of the present application. In the absence of conflict, the embodiment of the present application and the technical features in the embodiment can be combined with each other.

[0016] Figure 1 This is a flowchart of a tunnel lighting intelligent control method based on vehicle flow monitoring according to an embodiment of the present application, which is applied to a tunnel lighting intelligent control system and includes S101-S105.

[0017] S101: The tunnel lighting intelligent control system obtains the three-dimensional spatial information of the target tunnel to be matched with the lighting control strategy, the vehicle flow sensor monitoring information and the equipment network status information of the vehicle flow monitoring equipment network corresponding to the target tunnel; wherein the vehicle flow sensor monitoring information is the spatiotemporal domain monitoring information; the equipment network status information and the three-dimensional spatial information are non-spatiotemporal domain monitoring information.

[0018] S102: The intelligent control system for tunnel lighting determines, based on the vehicle flow sensor monitoring information, a first correlation degree that the target tunnel matches a plurality of initial lighting control strategies respectively through a debugged first control strategy generation network.

[0019] S103: The tunnel lighting intelligent control system identifies at least one tunnel section space vector of the three-dimensional space information, and identifies at least one device network state vector of the device network state information.

[0020] S104: The tunnel lighting intelligent control system generates a network through the debugged second control strategy, determines the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector, and separately determines the lighting intelligent control strategy vector of each lighting control decision vector, and performs full connection processing on each lighting control decision vector according to the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy vector of each lighting control decision vector, and determines the second correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively; wherein, the lighting control decision vector includes the at least one tunnel section space vector and the at least one device network state vector.

[0021] S105: The tunnel lighting intelligent control system determines a target lighting control strategy for the target tunnel by using a first correlation between the target tunnel and a plurality of initial lighting control strategies and a second correlation between the target tunnel and a plurality of the initial lighting control strategies.

[0022] First, in S101, the tunnel lighting intelligent control system obtains a series of key information. For the target tunnel, its three-dimensional spatial information is obtained, which covers non-temporal and spatial domain monitoring information such as the length, width, height of the tunnel and the layout of each area. For example, a tunnel is 2,000 meters long, 10 meters wide, and 7 meters high, and the entrance section, transition section, middle section, and exit section have different spatial layout characteristics. At the same time, the traffic flow sensor monitoring information of the traffic flow monitoring equipment network corresponding to the target tunnel is also obtained. The traffic flow sensor monitoring information here is spatiotemporal domain monitoring information, which can reflect the traffic flow situation at different times (such as every 5 minutes as a time interval) and different spatial locations (such as the tunnel entrance, different locations in the middle, etc.) in real time. For example, during peak hours, the number of vehicles passing through the tunnel entrance every 5 minutes may reach 200, while during low hours, there may be only 10 vehicles at the same location every 5 minutes. In addition, the device network status information is also obtained, including the working status of each sensor in the vehicle flow monitoring device network (such as microwave radar sensors or video detection sensors) (whether it is operating normally, whether there is a fault, etc.) and the status of the data transmission link (transmission rate, whether there is data packet loss, etc.) and other non-time and space domain monitoring information.

[0023] Next, in S102, the tunnel lighting intelligent control system determines the first correlation between the target tunnel and the multiple initial lighting control strategies respectively through the debugged first control strategy generation network based on the previously acquired vehicle flow sensor monitoring information. The initial lighting control strategy here may include multiple preset schemes. For example, one initial lighting control strategy is to set the lighting brightness of the tunnel entrance section and the middle section to the highest brightness level (such as 100% brightness) when the vehicle flow is greater than 150 vehicles / 5 minutes, and another may be to reduce the brightness of the middle section to 50% brightness level when the vehicle flow is less than 50 vehicles / 5 minutes. The first control strategy generation network may be debugged based on a large amount of historical data and simulation data, such as debugging by analyzing the effect data of different lighting control strategies under different vehicle flow conditions in the past month. For example, a certain initial lighting control strategy is to adjust the lighting brightness according to a specific vehicle flow threshold. When the vehicle flow sensor monitoring information shows that the current vehicle flow is in the vehicle flow interval set by the strategy, the first correlation between the strategy and the target tunnel will be high.

[0024] Then, in S103, the tunnel lighting intelligent control system starts to perform vector recognition on the acquired information. For three-dimensional spatial information, at least one tunnel section spatial vector is identified. For example, the spatial characteristics of the tunnel entrance section (such as a length of 100 meters, a width of 10 meters, a height of 7 meters, a specific lighting fixture layout, etc.) are quantified into a tunnel section spatial vector. For the device network status information, at least one device network state vector is identified. If there are 5 sensors in the vehicle flow monitoring device network, the states such as sensor 1 working normally and sensor 2 data transmission occasionally losing packets can be quantified into a device network state vector.

[0025] In S104, the tunnel lighting intelligent control system carries out further operations through the debugged second control strategy generation network. First, the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector is determined, where the lighting control decision vector includes at least one tunnel section space vector and at least one device network state vector previously identified. For example, the space vector of the tunnel entrance section and the corresponding device network state vector (such as the entrance section sensor is working properly) are fused. For example, there is a preset weight distribution mechanism in the second control strategy generation network, and different weights are assigned according to the importance of different vector elements. Then, the lighting intelligent control strategy vectors of each lighting control decision vector are determined respectively, and each lighting control decision vector is fully connected according to the lighting intelligent control strategy weighted vector of the fusion result, so as to determine the second correlation of the target tunnel matching multiple initial lighting control strategies respectively. For example, after a lighting control decision vector is weighted and fully connected, the matching degree with a certain initial lighting control strategy is quantified.

[0026] Finally, in S105, the tunnel lighting intelligent control system uses the first correlation of the target tunnel matching multiple initial lighting control strategies and the second correlation of the target tunnel matching multiple initial lighting control strategies to determine the target lighting control strategy of the target tunnel. For example, if a certain initial lighting control strategy scores high in the first correlation calculation and also scores high in the second correlation calculation, then the strategy is likely to be determined as the target lighting control strategy. For example, when calculating the first correlation, a certain strategy scores 0.8 (full score 1 point) and scores 0.75 when calculating the second correlation. After comprehensive consideration, it may become the final target lighting control strategy for intelligent control of the lighting of the target tunnel. This target lighting control strategy will guide the tunnel lighting intelligent control system to reasonably adjust the lighting brightness of each area in the tunnel according to multiple factors such as vehicle flow, tunnel space characteristics, and equipment network status, so as to achieve multiple goals such as energy saving, ensuring driving safety, and extending the life of lamps.

[0027] Through this series of operations, the intelligent control system of tunnel lighting obtains key information, generates network calculation correlation through different control strategies, and finally determines the target lighting control strategy, thus realizing intelligent, accurate and efficient control of the target tunnel lighting. This control method comprehensively considers multiple factors and is more scientific and reasonable than the traditional single factor control method. In practical applications, it can flexibly adjust the lighting strategy according to the actual situation of the tunnel, such as different traffic flow change patterns, unique tunnel space structure and the actual status of the equipment network, thereby improving the performance of the entire tunnel lighting system.

[0028] In the whole process, the first control strategy generation network and the second control strategy generation network play a key role. The first control strategy generation network mainly focuses on preliminarily evaluating the matching degree between the target tunnel and the initial lighting control strategy based on the traffic flow sensor monitoring information. For example, it may be based on a model based on traffic flow threshold division. When the traffic flow is in different intervals, the matching degree calculation method of the initial lighting control strategy is different. The second control strategy generation network pays more attention to the integration of multiple factors, integrating factors such as the tunnel section space vector and the device network state vector, and determining the correlation between the target tunnel and the initial lighting control strategy from a more comprehensive perspective.

[0029] For the tunnel section space vector, the elements it contains are not just simple spatial dimension information, but may also include factors such as the distribution density of lighting fixtures in different areas, the impact of special structures in different areas (such as curves, slopes, etc.) on light propagation, etc. Take curves as an example. Special lighting adjustments may be required at curves because the driver's field of vision changes when the vehicle is driving on the curve, which requires the lighting system to provide appropriate brightness to ensure driving safety. The device network state vector also contains rich content, such as sensor accuracy, data transmission stability, etc. If the accuracy of the sensor decreases, it may affect the accuracy of vehicle flow monitoring, and thus affect the selection of lighting control strategies.

[0030] When determining the target lighting control strategy, it is not a simple weighted average of the first and second relevances. Instead, it is based on the actual situation. For example, when the influence of traffic flow sensor monitoring information on lighting control is dominant, the first relevance may be given a higher weight; and when the special spatial structure of the tunnel or the equipment network status is abnormal, the weight of the second relevance may be increased accordingly. This flexible weight distribution mechanism enables the tunnel lighting intelligent control system to better adapt to various complex actual situations and ensure that the selected target lighting control strategy is the most suitable for the current situation of the target tunnel.

[0031] In actual tunnel lighting control, the implementation of the target lighting control strategy involves many aspects. For the brightness adjustment of lighting fixtures, there may be different adjustment methods according to different strategies. Taking LED lamps as an example, under the determined target lighting control strategy, if the traffic volume is low and there is no special spatial structure requirement in a certain area of ​​the tunnel, the brightness of the lamps may be adjusted to a lower level, such as reducing the brightness from the normal 80% to 30%. If the traffic volume is large or in key areas such as the entrance section of the tunnel, the lamps will be adjusted to a higher brightness, such as 90% or even 100%. This brightness adjustment must not only consider the current traffic volume and the spatial structure of the tunnel, but also the status of the equipment network. If the sensors in the equipment network show a sudden increase in traffic volume and the equipment network is in normal condition, the lighting system will quickly adjust the brightness of the lamps according to the target lighting control strategy to ensure driving safety.

[0032] In addition, the intelligent control system for tunnel lighting also needs to have a certain degree of adaptive capability. Over time, the traffic flow pattern of the tunnel may change. For example, due to the construction of new transportation facilities in the surrounding area or the adjustment of urban development plans, the period of originally low traffic flow may become a period of high traffic flow. At this time, the system needs to re-evaluate the relevance of each initial lighting control strategy based on the new traffic flow sensor monitoring information, and may adjust the target lighting control strategy. Similarly, if the tunnel is renovated or the equipment network is upgraded, the three-dimensional space information and equipment network status information have changed, and the system will re-perform the above series of operations to ensure that the lighting control strategy is always optimal.

[0033] From the perspective of energy saving, through the determination and implementation of this precise target lighting control strategy, the intelligent control system for tunnel lighting can effectively reduce energy consumption. In the traditional fixed lighting control mode, regardless of the volume of traffic, lighting fixtures often operate at a fixed brightness, which leads to energy waste during periods of low traffic. The system in the embodiment of the present application can dynamically adjust the brightness according to actual conditions, greatly reducing unnecessary energy consumption. For example, in a tunnel with obvious peaks and troughs in daily traffic, after adopting the system of the present invention, after a period of statistics, the average daily lighting energy consumption may be reduced by about 40% compared with the traditional mode.

[0034] From the perspective of driving safety, the target lighting control strategy can provide appropriate lighting brightness according to the actual traffic volume and tunnel conditions. In the case of heavy traffic, sufficient brightness can ensure that the driver can clearly see the road signs, other vehicles and pedestrians in the tunnel, reducing the probability of accidents. In special tunnel sections, such as the exit section, appropriate brightness adjustment can help the driver better adapt to the changes in light inside and outside the tunnel, avoiding accidents caused by visual discomfort caused by sudden changes in light.

[0035] In terms of equipment maintenance, since the system can adjust the lighting strategy according to the equipment network status information, when a sensor in the equipment network fails or the data transmission is unstable, the system can take corresponding measures, such as reducing the reliance on traffic flow monitoring data, and adjusting the lighting brightness to a relatively conservative but still safe level to avoid loss of control of lighting control due to equipment failure. At the same time, a reasonable lighting control strategy can also reduce the working time of lamps at unnecessary high brightness, thereby extending the service life of lamps and reducing equipment maintenance costs.

[0036] In summary, the intelligent control system for tunnel lighting improves the control level of tunnel lighting in many aspects through a series of operations in this invention application, achieves multiple goals such as energy saving, safety and equipment maintenance optimization, and provides a comprehensive and efficient solution for the intelligent management of tunnel lighting.

[0037] In actual large-scale tunnel networks, the advantages of this invention will be more obvious. For example, in an urban tunnel network containing multiple tunnels, each tunnel may have different traffic patterns, spatial structures, and equipment network states. The tunnel lighting intelligent control system in the embodiment of the present application can independently determine the target lighting control strategy for each tunnel to achieve refined management of the entire tunnel network. For tunnels with large traffic volume and complex structure, the system can accurately determine the appropriate lighting strategy based on its special circumstances; for tunnels with small traffic volume and relatively simple equipment network, it can also minimize energy consumption while ensuring safety.

[0038] In terms of data processing, the system needs to process a large amount of vehicle flow sensor monitoring information, three-dimensional space information, and equipment network status information. In order to ensure the efficient operation of the system, a distributed data processing architecture may be adopted. For example, in the vehicle flow monitoring equipment network, each sensor node can first perform preliminary processing and compression on the collected data, and then transmit the data to the central processing unit of the tunnel lighting intelligent control system. After receiving the data, the central processing unit can quickly perform the above series of operations to determine the target lighting control strategy. This distributed data processing method can reduce the amount of data transmission, improve the response speed of the system, and enable the lighting control to adapt to the actual situation in the tunnel more promptly.

[0039] The system also has certain advantages in dealing with emergencies. For example, when a traffic accident or other emergency occurs in a tunnel, the traffic volume may suddenly change. At this time, the traffic flow sensor monitoring information will be quickly fed back to the tunnel lighting intelligent control system, and the system can quickly adjust the lighting strategy according to the new situation. If the accident causes a sudden drop in traffic volume in a certain area, the system can reduce the lighting brightness in that area, on the one hand to avoid light interfering with rescue work, and on the other hand to reduce unnecessary energy consumption. If the accident causes vehicle congestion, the system can adjust the lighting according to the situation in the congested area and increase the brightness of the area to ensure the safety of rescuers and people in the car.

[0040] From a long-term development perspective, with the continuous advancement of technology, the intelligent control system for tunnel lighting can be further deeply integrated with other intelligent transportation systems. For example, data can be shared with the city's traffic command center, and the lighting strategy of the tunnel can be adjusted in advance based on the traffic flow forecast information of the entire city. If a large number of vehicles are predicted to enter the tunnel in a certain time period, the system can adjust the lighting brightness to an appropriate level in advance to improve the traffic efficiency and safety of the tunnel. At the same time, with the development of sensor technology, the accuracy and reliability of sensors in the vehicle flow monitoring equipment network will continue to improve, the acquisition of three-dimensional spatial information will also be more accurate, and the monitoring of equipment network status information will be more comprehensive, which will further enhance the performance of the intelligent control system for tunnel lighting in the embodiment of the present application, enabling it to better adapt to the complex and changeable tunnel lighting needs in the future.

[0041] In addition, with the continuous development of lighting fixture technology, tunnel lighting intelligent control systems also need to constantly adapt to new lamp characteristics. For example, with the continuous development of intelligent LED lamps, their dimming performance, color adjustment capabilities, etc. are constantly improving. The system in the embodiment of the present application can further optimize the lighting control strategy based on these new lamp characteristics. For example, under special weather conditions (such as foggy days), in addition to adjusting the brightness, the color of the lamp can also be adjusted to improve the penetration of light, thereby better ensuring driving safety. At the same time, as the life of the lamp is further extended and energy consumption is further reduced, the system can adjust the lighting strategy more flexibly to achieve greater energy saving while meeting safety and comfort.

[0042] In summary, the intelligent control system for tunnel lighting is based on the technical solution in the embodiment of the present application. By acquiring, analyzing and processing various information of the target tunnel, the target lighting control strategy is determined. It has demonstrated great advantages and potential in all aspects of tunnel lighting management, and provides solid technical support for the intelligent and efficient development of tunnel lighting.

[0043] In practical applications, the intelligent control system for tunnel lighting also needs to take into account the handling of some special situations. For example, during maintenance work in the tunnel, the lighting requirements are different from those during normal driving. At this time, the system can adjust the lighting strategy according to the area and nature of the maintenance work. If small-scale equipment maintenance is carried out in the middle section of the tunnel, the system can adjust the lighting brightness of the area to a level that meets the work needs of the maintenance personnel while avoiding excessive impact on the lighting of other normal driving areas. This may involve accurate identification of the maintenance area in the three-dimensional spatial information, and reasonable lighting adjustments based on the association between the maintenance equipment and the lighting system in the equipment network status information.

[0044] The system in the embodiment of the present application can also effectively manage special lighting needs in tunnels, such as emergency lighting and indicator lighting. In emergency situations (such as fire, earthquake, etc.), the system can ensure the normal operation of emergency lighting fixtures and provide sufficient brightness to guide vehicles and personnel to evacuate according to the pre-set emergency lighting strategy, combined with the current vehicle flow sensor monitoring information and equipment network status information. For example, when the vehicle flow monitoring equipment network detects vehicle congestion in a certain direction, the emergency lighting can strengthen the brightness guidance in that direction, and at the same time ensure the normal power supply and data transmission of the emergency lighting system according to the equipment network status information.

[0045] From the perspective of system reliability, the intelligent control system of tunnel lighting needs to have a certain fault tolerance. In the event of abnormalities in the traffic flow sensor monitoring information (such as interference with the sensor or data transmission errors), the system cannot completely lose the ability to control lighting. It can be estimated and adjusted based on historical traffic flow data and current three-dimensional spatial information and equipment network status information. For example, if the traffic flow sensor suddenly fails to work properly, the system can temporarily adjust the lighting strategy based on the average traffic flow in the same time period in the recent period and the current spatial structure and equipment network status of the tunnel to ensure the basic operation of the lighting system, and at the same time issue an alarm to notify relevant personnel to repair the sensor.

[0046] In terms of interaction with other related systems, in addition to integration with the intelligent transportation system, the tunnel lighting intelligent control system can also interact with the tunnel's ventilation system, monitoring system, etc. For example, when the ventilation system detects a decrease in air quality in the tunnel (such as increased smoke concentration), the lighting system can adjust the lighting strategy based on this information, such as increasing the brightness to facilitate the monitoring system to better observe the situation in the tunnel. At the same time, when the monitoring system detects abnormal behavior in the tunnel (such as pedestrians breaking in), the lighting system can adjust the lighting in the area where the abnormal behavior occurs through interaction with the monitoring system, such as increasing the brightness of the area, so as to better monitor and handle the situation.

[0047] In terms of updating and optimizing lighting control strategies, the intelligent control system for tunnel lighting can be continuously improved based on actual operation data. For example, by collecting data on vehicle flow and lighting control effects in different seasons and weather conditions, the first control strategy generation network and the second control strategy generation network can be re-debugged. If it is found that an initial lighting control strategy is not effective in a specific season or weather condition, the system can adjust or redesign the strategy to improve the adaptability and performance of the entire system.

[0048] To summarize, the intelligent control system for tunnel lighting is based on the technical solution of the embodiment of the present application. By comprehensively considering various information of the target tunnel, in different application scenarios, from normal driving lighting to special situation processing, from interaction with other systems to its own reliability assurance, and continuous optimization of lighting control strategies, it realizes efficient, intelligent, safe and energy-saving control of tunnel lighting, and provides a comprehensive lighting solution for tunnel operation and management.

[0049] In actual engineering implementation, the installation and debugging of the tunnel lighting intelligent control system needs to follow a certain process. First of all, during the tunnel construction or renovation stage, it is necessary to reasonably arrange the sensors in the traffic flow monitoring equipment network to ensure that the traffic flow sensor monitoring information can be accurately obtained. For example, the number and installation location of sensors are determined based on factors such as the length of the tunnel, the number of curves, and the location of the entrance and exit. For longer tunnels, it may be necessary to install sensors at the tunnel entrance, at certain intervals in the middle, and at the exit to fully monitor the traffic flow.

[0050] In order to obtain three-dimensional spatial information, accurate measurement and modeling are required. Advanced means such as laser scanning technology can be used to scan the internal structure of the tunnel in detail to obtain accurate length, width, height and special structural information of each area, so as to build an accurate three-dimensional spatial model. For the device network status information, during the system installation process, each device (such as sensors, data transmission equipment, etc.) needs to be initialized and tested to ensure that the status information of the device can be accurately obtained.

[0051] During the debugging phase, a large amount of historical data and simulation data are needed to train and optimize the first control strategy generation network and the second control strategy generation network. For example, traffic flow data, lighting control data, and related three-dimensional space and equipment network status data of other similar tunnels are collected and input into the network for training. At the same time, according to the actual test results, the parameters in the network are adjusted, such as weight allocation, threshold setting, etc., to ensure that the system can accurately determine the target lighting control strategy.

[0052] In an optional embodiment, the second control strategy generation network after debugging determines the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector, and separately determines the lighting intelligent control strategy vector of each lighting control decision vector, and performs full connection processing on each lighting control decision vector based on the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy vector of each lighting control decision vector, including: determining the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector based on the first network weight of the second control strategy generation network after debugging; determining the lighting intelligent control strategy weighted vector between the first network weight and the mapping eigenvalue of each lighting control decision vector; and determining the global decision vector of any two lighting control decision vectors between each lighting control decision vector based on the comparison result between the lighting intelligent control strategy vector of the fusion result and the determined lighting intelligent control strategy weighted vector.

[0053] The tunnel lighting intelligent control system follows a rigorous technical solution when performing related operations based on the debugged second control strategy generation network. First, the first network weight of the second control strategy generation network plays a key starting role in the whole process. This first network weight is determined by pre-debugging. It is a set of multiple weight values, each of which corresponds to a different element or factor related to the lighting control decision vector.

[0054] Taking a tunnel as an example, the lighting control decision vector contains space-related elements such as different sections of the tunnel (such as the entrance section, the middle section, and the exit section), equipment network status-related elements, etc. For the length factor in the space-related element of the entrance section, the first network weight may assign it a weight value of 0.2; for the sensor data transmission stability factor in the equipment network status-related element, a weight value of 0.15 may be assigned, etc. In this way, each lighting control decision vector is calculated according to the first network weight, so as to determine the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector. This fusion result lighting intelligent control strategy vector is a vector that comprehensively reflects the relationship between each lighting control decision vector and its role in the overall lighting control.

[0055] Next, determine the lighting intelligent control strategy weighted vector between the first network weight and the mapping eigenvalues ​​of each lighting control decision vector. The mapping eigenvalues ​​of the lighting control decision vector are calculated based on their own characteristics and their association with other relevant factors. For example, for the lighting control decision vector of the middle section, its mapping eigenvalue may be related to the expected traffic flow, lamp layout density, and special structures (such as curves, slopes, etc.) of the section. For example, when the expected traffic flow in the middle section is large, the mapping eigenvalue related to the traffic flow may be 0.6; when the lamp layout density is moderate, the corresponding mapping eigenvalue may be 0.3; when the special structure has a small impact, the relevant mapping eigenvalue may be 0.1. Then multiply the corresponding weight values ​​in the first network weights by these mapping eigenvalues ​​and add them up to obtain the lighting intelligent control strategy weighted vector. For example, if the weight value related to traffic flow in the first network weight is 0.2, the weight value related to the lamp layout density is 0.3, and the weight value related to the special structure is 0.1, then the part related to traffic flow in the lighting intelligent control strategy weighted vector is 0.2×0.6=0.12, the part related to the lamp layout density is 0.3×0.3=0.09, and the part related to the special structure is 0.1×0.1=0.01. After accumulation, the lighting intelligent control strategy weighted vector is obtained.

[0056] Afterwards, the global decision vector of any two lighting control decision vectors between the lighting control decision vectors is determined based on the comparison results between the lighting intelligent control strategy vector of the fusion result and the determined lighting intelligent control strategy weighted vector. This comparison process involves comparative analysis of the corresponding elements of the two vectors. For example, an element in the lighting intelligent control strategy vector of the fusion result indicates that the recommended brightness level of the entrance section lighting under a specific traffic flow is 80%, while the recommended brightness level calculated by the corresponding element in the lighting intelligent control strategy weighted vector is 70%. This difference will be taken into account in the comparison results. When determining the global decision vector of any two lighting control decision vectors, the impact of this comparison result is multifaceted. For example, there are lighting control decision vectors A and B. When considering the relevant factors of the entrance section, by comparing the comparison results of the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy weighted vector, it is found that some elements of vector A in the entrance section are more matched with the overall situation, while vector B may have better performance in other sections. Then, when constructing the global decision vector, it is necessary to comprehensively consider these factors. It is possible to assign a higher weight to the A vector in the entrance section and assign corresponding weights to the B vector in other sections. Finally, a global decision vector that reflects the overall relationship and importance of the A and B vectors is obtained.

[0057] This technical solution is of great significance in the intelligent control system of tunnel lighting. From the perspective of determining the lighting control strategy, this method can more accurately consider the impact of various factors on the final lighting control decision. For example, under different traffic conditions, the lighting system can accurately adjust the brightness of different sections through the precise calculation of the first network weight and the mapping eigenvalue. If the traffic volume is at a peak during a period of time, the traditional rough control method may simply adjust the brightness of all sections to the highest, but through this technical solution, more detailed adjustments can be made according to the actual conditions of different sections (such as the special importance of the entrance section, the traffic flow distribution in the middle section, etc.), such as adjusting the brightness of the entrance section to 100%, and adjusting the brightness of the middle section to 80%-90% according to the different areas of traffic flow density, etc.

[0058] From the perspective of adapting to different tunnel environments, the spatial structures and equipment network states of different tunnels vary greatly. Through this technical solution, effective lighting control can be carried out according to the characteristics of different tunnels. For example, a long tunnel and a short tunnel, the long tunnel may have more intermediate sections that require special consideration, while the short tunnel may pay more attention to the lighting of the entrance and exit sections. The first network weights and mapping eigenvalues ​​in this technical solution can be adjusted according to the actual conditions of different tunnels, so as to determine the lighting control decision vectors suitable for each tunnel, and then obtain a reasonable global decision vector to achieve the optimal lighting control strategy.

[0059] From the perspective of system flexibility and scalability, as the operation of the tunnel develops, new situations may arise, such as the addition of new equipment, long-term changes in traffic patterns, etc. This technical solution can easily adjust the first network weights and mapping eigenvalues ​​to adapt to these changes. For example, when a new energy-saving lighting device is added to the tunnel, its lighting characteristics are different from those of the original device. By adjusting the mapping eigenvalues ​​related to the device, the system can re-evaluate the lighting control decision vector, and then adjust the global decision vector, so as to achieve effective integration of the new device and optimization of the lighting control strategy.

[0060] In this way, through a series of rigorous vector calculation and comparison processes, a more accurate, flexible and adaptable lighting control strategy determination method for various environments is realized in the tunnel lighting intelligent control system. Accuracy is reflected in the ability to accurately determine the relationship between lighting control decision vectors based on the weights and eigenvalues ​​of different factors, as shown in the numerical examples of brightness adjustment under different traffic flows and tunnel structures. Flexibility and adaptability are reflected in the ability to adjust relevant parameters according to different tunnel conditions (such as the difference between long and short tunnels) and changes in operation (such as the introduction of new equipment), ensuring that the lighting control strategy is always in the optimal state, improving the overall performance of the system and its adaptability to different tunnel environments.

[0061] In a preferred design concept, the target lighting control strategy of the target tunnel is determined by utilizing the first correlation that the target tunnel is matched with multiple initial lighting control strategies and the second correlation that the target tunnel is matched with multiple initial lighting control strategies, including: determining the third correlation that the target tunnel is matched with multiple initial lighting control strategies according to the first bias item coefficient of the first control strategy generation network and the second bias item coefficient of the second control strategy generation network, the first correlation that the target tunnel is matched with multiple initial lighting control strategies, and the second correlation that the target tunnel is matched with multiple initial lighting control strategies; and taking the initial lighting control strategy corresponding to the largest third correlation as the target lighting control strategy of the target tunnel.

[0062] The tunnel lighting intelligent control system operates based on a preferred design idea when determining the target lighting control strategy for the target tunnel. This process involves the comprehensive use of the first bias term coefficient of the first control strategy generation network and the second bias term coefficient of the second control strategy generation network, and the first correlation and second correlation of the target tunnel matching multiple initial lighting control strategies.

[0063] First, the first bias term coefficient and the second bias term coefficient are important parameters pre-set in their respective control strategy generation networks. The first bias term coefficient has a specific influence on the matching relationship between the target tunnel and the initial lighting control strategy in the first control strategy generation network. For example, the first bias term coefficient may be a specific value, such as 0.3, which adjusts the role of the first correlation in the final determination of the target lighting control strategy to a certain extent. Similarly, the second bias term coefficient also plays a similar role in the second control strategy generation network, for example, its value is 0.2.

[0064] When determining the third correlation of the target tunnel matching multiple initial lighting control strategies, these factors need to be considered comprehensively. Taking a certain initial lighting control strategy as an example, the first correlation is 0.6, the second correlation is 0.5, the first bias term coefficient is 0.3, and the second bias term coefficient is 0.2. Then the process of calculating the third correlation is as follows: the third correlation is a comprehensive correlation obtained by combining the first correlation and the second correlation with their respective bias term coefficients. The specific calculation method is to multiply the first correlation by the first bias term coefficient, and then multiply the second correlation by the second bias term coefficient through a specific calculation logic, and then combine the two results in some way (for example, adding or adding according to a specific weight, etc.). Calculated by adding, the third correlation of the initial lighting control strategy is (0.6×0.3+0.5×0.2)=0.28.

[0065] This process is repeated for each initial lighting control strategy. For another initial lighting control strategy, for example, its first correlation is 0.5, and its second correlation is 0.6. According to the same calculation method, its third correlation is (0.5×0.3+0.6×0.2)=0.27.

[0066] After completing the calculation of the third correlations of all initial lighting control strategies, the system will use the initial lighting control strategy corresponding to the largest third correlation as the target lighting control strategy for the target tunnel. For example, if the third correlations of three initial lighting control strategies are calculated to be 0.28, 0.27, and 0.25, respectively, then the initial lighting control strategy corresponding to the third correlation of 0.28 will be determined as the target lighting control strategy.

[0067] This method of determining the target lighting control strategy is of great significance in the intelligent control system of tunnel lighting. From the perspective of accuracy, by introducing the bias term coefficient, the weights of the first correlation and the second correlation in the final decision can be adjusted more accurately. For example, in some cases, the vehicle flow sensor monitoring information may be more important. At this time, the influence of the first correlation in the calculation of the third correlation can be increased by adjusting the first bias term coefficient. For example, when the first bias term coefficient is increased from 0.3 to 0.4, for the initial lighting control strategy with a first correlation of 0.6 and a second correlation of 0.5, the third correlation becomes (0.6×0.4+0.5×0.2)=0.34, which indicates that the proportion of vehicle flow-related factors in the decision-making has increased, so that the final target lighting control strategy is more in line with actual needs.

[0068] From the perspective of adaptability, different tunnel environments and operating conditions may require different weight allocations. For example, for a tunnel with frequent changes in traffic flow and a simple tunnel structure, it may be desirable to rely more on traffic flow sensor monitoring information, that is, the first correlation has a greater impact; while for a tunnel with relatively stable traffic flow but a complex tunnel structure, it may be necessary to pay more attention to factors such as the tunnel section space vector and the equipment network state vector contained in the second correlation. By adjusting the bias term coefficient, the system can adapt to these different situations. For example, in a tunnel with stable traffic flow and a complex tunnel structure, reducing the first bias term coefficient to 0.2 and increasing the second bias term coefficient to 0.3 will give the second correlation more weight when calculating the third correlation, thereby selecting a target lighting control strategy that is more suitable for this tunnel environment.

[0069] From the perspective of system optimization, this method can be continuously adjusted as data accumulates during tunnel operation. Over time, if it is found that some initial lighting control strategies perform poorly in actual applications, it may be because the settings of the first bias item coefficient and the second bias item coefficient are unreasonable. By analyzing the actual lighting effect data, the bias item coefficient can be adjusted. For example, it is found that a certain initial lighting control strategy has the highest third correlation calculated according to the existing bias item coefficient in a certain time period and is selected as the target lighting control strategy, but the actual lighting effect does not meet the expected energy saving or safety standards. After analysis, it is found that the weight of vehicle flow information in this strategy is too high, resulting in unreasonable lighting control during periods with small vehicle flow but special tunnel space structure. Then the first bias item coefficient can be appropriately reduced, the second bias item coefficient can be increased, and the third correlation can be recalculated to optimize the selection of the target lighting control strategy.

[0070] In this way, by introducing the first bias item coefficient of the first control strategy generation network and the second bias item coefficient of the second control strategy generation network, the third correlation is determined in combination with the first correlation and the second correlation of the target tunnel matching multiple initial lighting control strategies, and then the target lighting control strategy is selected. This method improves the accuracy of determining the target lighting control strategy, can adjust the weight of each correlation according to different tunnel environments and operating conditions, enhances the adaptability of the system, and can continuously optimize the bias item coefficient based on actual data accumulation, thereby optimizing the selection of the target lighting control strategy and improving the overall performance of the tunnel lighting intelligent control system.

[0071] In another optional embodiment, the first control strategy generation network and the second control strategy generation network are debugged by the following steps: obtaining a training information set; the training information set includes three-dimensional space information samples corresponding to multiple tunnel samples, vehicle flow sensor monitoring information samples and equipment network status information samples of vehicle flow monitoring equipment network samples corresponding to each tunnel sample; each tunnel sample has a priori lighting control strategy; obtaining a debugging example set and a verification example set from the training information set; wherein the debugging example set and the verification example set do not contain data corresponding to the same tunnel samples; performing at least one debugging process on the original control strategy generation network according to the debugging example set, and verifying the original control strategy generation network obtained by each debugging process according to the verification example set, obtaining the verification result corresponding to each debugging process, until the set debugging termination requirements are met, and determining the debugged first control strategy generation network and the second control strategy generation network based on the verification result corresponding to each debugging process; wherein the original control strategy generation network includes a first original control strategy generation network corresponding to the first control strategy generation network and a second original control strategy generation network corresponding to the second control strategy generation network.

[0072] In a further embodiment, the debugging process includes: arbitrarily selecting a part of the first historical examples from the debugging example set as the historical examples corresponding to the current debugging process; determining that the tunnel samples match the first correlation debugging values ​​of multiple initial lighting control strategies respectively through the first original control strategy generation network based on the vehicle flow sensor monitoring information samples in the first historical examples; identifying the lighting control decision vector samples for the second original control strategy generation network in the first historical examples; the lighting control decision vector samples include the tunnel sample segment space vectors of multiple tunnel samples and the sample device network state vectors corresponding to the tunnel samples; based on the lighting control decision vector samples, Determine, through a second original control strategy generation network, that a tunnel sample matches a second correlation debugging value of a plurality of initial lighting control strategies respectively; determine a lighting control strategy debugging result of the tunnel sample based on that the tunnel sample matches a first correlation debugging value of a plurality of initial lighting control strategies respectively, and that the target tunnel matches a second correlation debugging value of a plurality of the initial lighting control strategies respectively; optimize the network weights of the original control strategy generation network based on the difference between the lighting control strategy debugging result and the prior lighting control strategy, and use the original control strategy generation network after optimizing the network weights as the original control strategy generation network before the next debugging process.

[0073] In this optional embodiment, debugging of the first control strategy generation network and the second control strategy generation network is a complex and orderly process.

[0074] The first step is to obtain a training information set, which contains rich content. It includes three-dimensional spatial information samples corresponding to multiple tunnel samples. For example, a three-dimensional spatial information sample of a tunnel sample may be a tunnel structure information with a length of 1,500 meters, a width of 9 meters, and a height of 6 meters, and with a specific entrance section, transition section, middle section, and exit section layout; traffic flow sensor monitoring information samples of the traffic flow monitoring equipment network samples corresponding to each tunnel sample, such as a certain tunnel sample during peak hours, the traffic flow monitoring equipment network sample shows that the number of vehicles passing through every 10 minutes is 300, and the number of vehicles passing through every 10 minutes during the valley period is 50; and equipment network status information samples, such as the working status of sensors in the equipment network (normal or with a certain proportion of failure risk), data transmission stability (such as a transmission packet loss rate of 1%, etc.). In addition, each tunnel sample has a priori lighting control strategy, which is the lighting control strategy for each tunnel sample determined based on experience or preliminary design.

[0075] Then, the debugging example set and the verification example set are obtained from the training information set. There is a clear requirement for these two example sets, that is, they do not contain data corresponding to the same tunnel samples. This division is to ensure the accuracy and effectiveness of the debugging and verification process.

[0076] Next, a debugging process is performed. In each debugging process, a portion of the first historical examples are randomly selected from the debugging example set and used as the historical examples corresponding to the current debugging process. Based on the vehicle flow sensor monitoring information sample in the first historical example, the first original control strategy generation network is used to determine the first correlation debugging values ​​that the tunnel sample matches with the multiple initial lighting control strategies. For example, there are three initial lighting control strategies. For a certain tunnel sample, the first correlation debugging values ​​calculated by the first original control strategy generation network based on the vehicle flow sensor monitoring information sample are 0.4, 0.5, and 0.3 respectively.

[0077] At the same time, the lighting control decision vector samples for the second original control strategy generation network in the first historical example are identified, and these lighting control decision vector samples include tunnel sample segment space vectors of multiple tunnel samples and sample device network state vectors corresponding to the tunnel samples. Based on the lighting control decision vector samples, the second original control strategy generation network is used to determine the second correlation debugging values ​​that the tunnel samples match with the multiple initial lighting control strategies respectively. For example, for the above tunnel samples, the second correlation debugging values ​​calculated by the second original control strategy generation network are 0.3, 0.4 and 0.2 respectively.

[0078] Afterwards, according to the first correlation debugging values ​​of the tunnel sample that are respectively matched to the multiple initial lighting control strategies, and the second correlation debugging values ​​of the target tunnel that are respectively matched to the multiple initial lighting control strategies, the lighting control strategy debugging result of the tunnel sample is determined. For example, by combining the first correlation debugging value and the second correlation debugging value, a comprehensive debugging result value is obtained for a certain initial lighting control strategy.

[0079] Finally, based on the difference between the lighting control strategy debugging results and the prior lighting control strategy, the network weights of the original control strategy generation network are optimized. If there are large differences between the lighting control strategy debugging results and the prior lighting control strategy in certain key indicators (such as the brightness adjustment strategy for a specific traffic flow), the network weights in the original control strategy generation network will be adjusted. The original control strategy generation network after optimizing the network weights is used as the original control strategy generation network before the next debugging process. This process is repeated continuously, and the original control strategy generation network obtained from each debugging process is verified based on the verification example set to obtain the verification results corresponding to each debugging process until the set debugging termination requirements are met. The debugged first control strategy generation network and the second control strategy generation network are determined based on the verification results corresponding to each debugging process.

[0080] This debugging method is of great significance in many aspects of the entire tunnel lighting intelligent control system. From the perspective of improving network accuracy, debugging through a large amount of tunnel sample data enables the first control strategy generation network and the second control strategy generation network to more accurately determine the relevance based on different input information. For example, after multiple debuggings, the network can more accurately judge the degree of match with the initial lighting control strategy for different vehicle flows and tunnel structures. When vehicle flow monitoring information and tunnel space and equipment network status information are input, a relevance result that is more in line with actual needs can be output, thereby improving the accuracy of determining the final target lighting control strategy.

[0081] In terms of network adaptability, the use of multiple different tunnel sample data covers a variety of possible tunnel structures, traffic patterns, and device network states, which enables the debugged network to adapt to different actual tunnel conditions. For example, whether it is a long tunnel or a short tunnel, a tunnel with large or small traffic volume, the debugged network can effectively handle and provide appropriate lighting control strategy relevance.

[0082] In terms of optimization, the network can continuously improve itself by optimizing the network weights based on the difference between the lighting control strategy debugging results and the prior lighting control strategy. If it is found during the debugging process that a certain network weight setting leads to deviations from the prior lighting control strategy in terms of energy saving or safety lighting, the weights can be adjusted in time to make the network develop in a better direction. Moreover, by verifying the verification example set, the effectiveness of the network on different data subsets can be guaranteed, and problems such as overfitting can be avoided, thereby further improving the performance of the network and ensuring that in practical applications, it can accurately provide appropriate control strategy correlation calculations for the tunnel lighting intelligent control system, thereby determining the optimal target lighting control strategy.

[0083] In this way, the first control strategy generation network and the second control strategy generation network are determined by dividing the debugging example set and the verification example set based on the training information set containing multiple tunnel sample information, performing multiple debugging processes and verifications. This method uses a large amount of actual data to improve the accuracy of the network, can adapt to different tunnel conditions and enhance adaptability, optimizes the network weights based on the difference between the debugging results and the prior strategies, and ensures the effectiveness and accuracy of the network in determining the relevance of the lighting control strategy, thereby improving the performance of the entire tunnel lighting intelligent control system.

[0084] Under another optional design idea, the second correlation debugging values ​​of the tunnel samples respectively matching the multiple initial lighting control strategies are determined through the second original control strategy generation network based on the lighting control decision vector samples, including: determining the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector sample through the second original control strategy generation network, and determining the lighting intelligent control strategy vector of each lighting control decision vector sample respectively, and performing full connection processing on each lighting control decision vector sample according to the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy vector of each lighting control decision vector, and determining the second correlation debugging values ​​of the tunnel samples respectively matching the multiple initial lighting control strategies.

[0085] Under this optional design idea, the process of determining that the tunnel samples are matched to the second correlation debugging values ​​of multiple initial lighting control strategies respectively through the second original control strategy generation network based on the lighting control decision vector samples is a complex and sophisticated operation process.

[0086] First, the lighting control decision vector sample contains important elements such as the tunnel sample segment space vector of multiple tunnel samples and the sample device network state vector corresponding to the tunnel sample. In this process, the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector sample is determined by the second original control strategy generation network. For example, there are three lighting control decision vector samples, namely vector sample A, vector sample B and vector sample C. Vector sample A contains the spatial vector information of the entrance section of a specific tunnel sample (such as relevant spatial parameters such as length, width, height, and information such as the layout of lighting fixtures in the area) and the corresponding sample device network state vector information (such as the working status of the entrance section sensor, data transmission stability, etc.); vector sample B contains relevant information of the middle section; vector sample C contains relevant information of the exit section.

[0087] The second original control strategy generation network has its specific structure and parameter settings, which will process each lighting control decision vector sample. When determining the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector sample, it is necessary to comprehensively consider the relationship between each element and their role in the overall lighting control strategy. For example, a certain spatial parameter (such as length) of the entry segment space vector in vector sample A may be assigned a certain weight, and the sensor working state in the device network state vector is also assigned a specific weight. Through the allocation of these weights and the calculation logic in the network, a component value of vector sample A in the fusion process is obtained. Vector sample B and vector sample C are processed in the same way, and finally the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector sample is obtained.

[0088] At the same time, the lighting intelligent control strategy vector of each lighting control decision vector sample must be determined separately. This process also depends on the parameters and calculation logic in the second original control strategy generation network. For each lighting control decision vector sample, calculations are performed based on its own element characteristics and the pre-set rules in the network. For example, the lighting intelligent control strategy vector of vector sample A may be determined based on a specific combination of the spatial vector of the entry segment and the device network state vector, and a vector reflecting the characteristics of the vector sample in terms of lighting control strategy is obtained through the mapping relationship in the network.

[0089] Next, the lighting intelligent control strategy vectors of each lighting control decision vector are operated according to the lighting intelligent control strategy weighted vector of the fusion result, and each lighting control decision vector sample is fully connected. This lighting intelligent control strategy weighted vector is determined based on the lighting intelligent control strategy vector of the fusion result and other related factors. For example, the lighting intelligent control strategy vector of the fusion result has specific values ​​on certain key elements (such as elements related to brightness adjustment), which will affect the formation of the lighting intelligent control strategy weighted vector. In the process of full connection processing, the lighting intelligent control strategy weighted vector and the lighting intelligent control strategy vector of each lighting control decision vector are interactively calculated.

[0090] For example, for the lighting intelligent control strategy vector of vector sample A, some weight values ​​in the lighting intelligent control strategy weight vector will be multiplied with the corresponding elements in the lighting intelligent control strategy vector of vector sample A, and then these product results will be accumulated or other combined operations will be performed according to the rules of full connection processing. The same operation is performed on vector samples B and vector samples C.

[0091] Through such a series of operations, it is finally determined that the tunnel sample matches the second correlation adjustment values ​​of multiple initial lighting control strategies. For example, there are three initial lighting control strategies, namely strategy 1, strategy 2 and strategy 3. After the above complex calculation process, for the tunnel sample, a second correlation adjustment value of 0.4 is obtained for strategy 1, a second correlation adjustment value of 0.5 is obtained for strategy 2, and a second correlation adjustment value of 0.3 is obtained for strategy 3.

[0092] This method of determining the second relevance debugging value is of great significance in the entire tunnel lighting intelligent control system. From the perspective of data processing accuracy, through the careful processing of the lighting control decision vector samples, including fusion, separate determination of the lighting intelligent control strategy vectors, and full connection processing, the influence of various factors of the tunnel sample on the matching degree of the initial lighting control strategy can be more accurately considered. For example, the specific conditions of the spatial vectors and device network state vectors in different sections can be accurately reflected in the second relevance debugging value. If the spatial structure of the middle section of a tunnel sample is complex and there are some unstable factors in the device network state, this situation will be reflected in the second relevance debugging value through the above processing process, so that the matching degree for different initial lighting control strategies can accurately reflect the actual situation of the tunnel sample.

[0093] From the perspective of network effectiveness, this approach enables the second original control strategy generation network to fully utilize the information in the lighting control decision vector samples and improve the network's ability to judge the relationship between different tunnel samples and the initial lighting control strategy. For example, as the number of tunnel samples increases and the content of the lighting control decision vector samples becomes richer, the network can more effectively determine the second correlation debugging value based on this information, thereby improving the effectiveness of the entire network in dealing with the problem of matching different tunnel lighting control strategies.

[0094] From the perspective of system optimization, since the second relevance debugging value can be accurately determined, this provides an important basis for the optimization of the entire tunnel lighting intelligent control system. For example, if it is found that the second relevance debugging value for a certain initial lighting control strategy is generally low in multiple tunnel samples, this may indicate that the initial lighting control strategy needs to be adjusted or optimized. At the same time, the second relevance debugging value determined in this way also helps to optimize the second original control strategy generation network itself, such as adjusting the weight distribution or calculation logic in the network to improve the accuracy and effectiveness of the network in determining the second relevance debugging value.

[0095] Therefore, the second correlation debugging value is determined by performing a series of complex vector processing operations based on the lighting control decision vector sample and the second original control strategy generation network. This method improves the accuracy of data processing and can accurately reflect the impact of tunnel sample factors on the matching degree of the initial lighting control strategy; enhances the effectiveness of the network, enabling the network to better use vector sample information to judge relationships; and provides a basis for the optimization of the system and the network itself, thereby improving the performance of the entire tunnel lighting intelligent control system.

[0096] Under another optional design idea, the first correlation debugging value that determines that the tunnel sample matches multiple initial lighting control strategies respectively is determined based on the vehicle flow sensor monitoring information sample in the first historical example through the first original control strategy generation network, including: determining a monitoring thermal mark value in the vehicle flow sensor monitoring information sample that is not less than a first set monitoring thermal intensity; wherein the first set monitoring thermal intensity is determined based on a set vehicle flow coefficient in the vehicle flow sensor monitoring information sample; cleaning out the monitoring thermal mark value in the vehicle flow sensor monitoring information sample to obtain a first vehicle flow sensor monitoring information sample, and determining a lighting control decision vector for the first original control strategy generation network based on the first vehicle flow sensor monitoring information sample.

[0097] Furthermore, the method of determining the lighting control decision vector for the first original control strategy generation network based on the first vehicle flow sensor monitoring information sample includes: determining a second vehicle flow sensor monitoring information sample in which the sensor monitoring thermal value in the first vehicle flow sensor monitoring information sample is not less than a second set monitoring thermal intensity; wherein the second set monitoring thermal intensity is less than the first set monitoring thermal intensity; updating the second vehicle flow sensor monitoring information sample to obtain an updated vehicle flow sensor monitoring sample; and determining the lighting control decision vector for the first original control strategy generation network by combining the updated vehicle flow sensor monitoring sample and the third vehicle flow sensor monitoring information sample; wherein the third vehicle flow sensor monitoring information sample is the vehicle flow sensor monitoring information sample in the first vehicle flow sensor monitoring information sample excluding the second vehicle flow sensor monitoring information sample.

[0098] Under this optional design idea, the entire process of determining whether the tunnel samples are matched to the first correlation debugging values ​​of multiple initial lighting control strategies has a rigorous and complex logical flow.

[0099] First, for the process of determining the first correlation debugging value through the first original control strategy generation network based on the vehicle flow sensor monitoring information sample in the first historical example, one of the important operations is to determine the monitoring thermal power mark value in the vehicle flow sensor monitoring information sample. The first set monitoring thermal power here plays a key defining role, and the first set monitoring thermal power is determined based on the set vehicle flow coefficient in the vehicle flow sensor monitoring information sample. For example, the vehicle flow sensor monitoring information sample contains a wealth of vehicle flow related data, such as the number of vehicle flows in different time periods (with each 15 minutes as a time period). The set vehicle flow coefficient may be a value obtained based on the statistical analysis of historical data. For example, after statistical analysis of the vehicle flow data in the past month, a coefficient value corresponding to the average vehicle flow level is obtained. Based on this coefficient value, the first set monitoring thermal power is determined to be a specific value, such as 60 (here is just a value set for the convenience of explanation). Then in the vehicle flow sensor monitoring information sample, those values ​​of the monitoring thermal power that are not less than 60 are marked as monitoring thermal power mark values. These monitored thermal signatures usually represent special conditions of traffic flow, which may be peak hours of traffic flow or abnormally high traffic flow due to special events (such as a surge in traffic flow during holidays).

[0100] Next, the monitoring thermal marker values ​​in the traffic flow sensor monitoring information sample are cleaned to obtain the first traffic flow sensor monitoring information sample. The purpose of this step is to remove those special data points that may have an excessive impact on subsequent analysis, so that the data is purer and can better reflect the general rules of traffic flow. For example, if the cleaning operation is not performed, the extremely high traffic flow data during peak hours may dominate the subsequent analysis results, causing the final lighting control decision vector to be biased towards dealing with high traffic flow situations, and not be well adapted to the lighting needs during normal traffic flow. The cleaned first traffic flow sensor monitoring information sample can provide a more accurate data basis for the subsequent determination of the lighting control decision vector.

[0101] Then, the lighting control decision vector for the first original control strategy generation network is determined based on the first vehicle flow sensor monitoring information sample. In this process, it is first necessary to determine the second vehicle flow sensor monitoring information sample in which the sensor monitoring thermal value in the first vehicle flow sensor monitoring information sample is not less than the second set monitoring thermal intensity. The second set monitoring thermal intensity here is less than the first set monitoring thermal intensity. For example, the second set monitoring thermal intensity is set to 30. In the first vehicle flow sensor monitoring information sample, find those parts whose sensor monitoring thermal values ​​are not less than 30 as the second vehicle flow sensor monitoring information sample. Although this part of the data is not as extreme as the monitoring thermal mark value that was cleaned up before, it still has a certain particularity or importance. For example, these data may represent a situation where the traffic volume is relatively high but not during peak hours.

[0102] Afterwards, the second vehicle flow sensor monitoring information sample is updated to obtain a vehicle flow sensor monitoring update sample. This update operation is to further optimize this part of the data to make it more in line with the actual situation or better reflect its importance in the lighting control decision. For example, for some data points in the second vehicle flow sensor monitoring information sample, it may be adjusted according to the change trend of the vehicle flow data in the adjacent time period. For example, the vehicle flow data of a certain time period is 40 in the original data (satisfying the second set monitoring intensity), but according to the steady upward trend of the vehicle flow in the previous and next time periods, this data can be updated to 42 to more accurately reflect the dynamic changes of the vehicle flow.

[0103] Finally, the lighting control decision vector of the first original control strategy generation network is determined by combining the vehicle flow sensor monitoring update sample and the third vehicle flow sensor monitoring information sample. Among them, the third vehicle flow sensor monitoring information sample is the vehicle flow sensor monitoring information sample in the first vehicle flow sensor monitoring information sample except the second vehicle flow sensor monitoring information sample. Through this combination, various situations in the first vehicle flow sensor monitoring information sample can be comprehensively considered. For example, the data in the vehicle flow sensor monitoring update sample may more reflect the situation where the vehicle flow is relatively high or has a special change trend, while the third vehicle flow sensor monitoring information sample more represents the general situation of the vehicle flow or the situation of low vehicle flow. Combining these two parts of data can form a lighting control decision vector that comprehensively considers different vehicle flow situations. This lighting control decision vector may contain preliminary decision information on lighting control for different sections of the tunnel (such as the entrance section, the middle section, and the exit section) at different vehicle flow levels. For example, for the entrance section, information such as a recommended value of lighting brightness when the traffic volume is low and another recommended value when the traffic volume is high may be included in the lighting control decision vector.

[0104] This method of determining the first correlation debugging value is of great significance in many aspects in the intelligent control system of tunnel lighting. From the perspective of data accuracy, by determining and cleaning the monitoring thermal marker value, the interference of extreme data on the overall data can be effectively avoided. For example, in a tunnel example, if such processing is not performed, the lighting control decision vector finally determined may tend to high-brightness lighting in all time periods due to extremely high traffic flow data (monitoring thermal marker value) during a few peak hours. After processing, the lighting control decision vector can be determined more accurately according to normal and different levels of traffic flow conditions, thereby improving the accuracy of data processing.

[0105] In terms of system adaptability, this approach can better adapt to complex changes in traffic flow. Since data with different monitoring intensities are taken into account and processed separately (the second traffic flow sensor monitoring information sample is determined and updated, and then combined with the third traffic flow sensor monitoring information sample), the system can make reasonable considerations when determining the lighting control decision vector, whether it is normal traffic flow fluctuations or traffic flow changes caused by special events. For example, for slight fluctuations in daily traffic flow and occasional peak traffic flow on holidays, the system can determine the appropriate lighting control decision vector based on this processing method, thereby improving the system's adaptability to different traffic flow patterns.

[0106] From the perspective of optimization, this method helps to improve the accuracy and effectiveness of the first original control strategy generation network in determining the first relevance debugging value. Through the careful processing of the vehicle flow sensor monitoring information samples, the lighting control decision vector can be determined more accurately, thereby improving the accuracy of the first relevance debugging value. If the first relevance debugging value can more accurately reflect the degree of match between the tunnel sample and the initial lighting control strategy, then the initial lighting control strategy can be more effectively adjusted and optimized, thereby improving the performance of the entire tunnel lighting intelligent control system. For example, if it is found that an initial lighting control strategy has a low degree of match with the tunnel sample after being evaluated based on the accurate first relevance debugging value, the strategy can be adjusted in a targeted manner, such as adjusting the brightness setting under different vehicle flows, etc., to improve the lighting control effect of the entire system.

[0107] In this way, by determining data of different monitoring heat intensities based on the vehicle flow sensor monitoring information sample and performing relevant processing to determine the lighting control decision vector and then determine the first correlation debugging value. This method improves the accuracy of data processing and avoids extreme data interference; enhances the system's adaptability to complex changes in vehicle flow; and improves the accuracy and effectiveness of the first original control strategy generation network in determining the first correlation debugging value, thereby optimizing the performance of the entire tunnel lighting intelligent control system.

[0108] In summary, the embodiment of the present application determines the lighting control strategy by obtaining various information of the target tunnel and performing a series of processing, and comprehensively considers three-dimensional spatial information, vehicle flow sensor monitoring information and equipment network status information, so that the determination of the lighting control strategy is more comprehensive. The correlation is determined by using the debugged network to improve the accuracy of strategy matching. Identifying different vectors and performing fusion processing can carefully consider the mutual influence of various factors in the tunnel to ensure that the lighting control strategy adapts to the actual conditions of the tunnel. This method can determine the target lighting control strategy according to different correlations, which helps to improve the intelligence of the lighting system, thereby achieving energy saving, ensuring driving safety and extending the service life of lamps, and improving the overall efficiency of tunnel lighting management.

[0109] An embodiment of the present application provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the intelligent control method for tunnel lighting based on vehicle flow monitoring.

[0110] An embodiment of the present application provides a processor, which is used to run a program, wherein the program executes the intelligent control method of tunnel lighting based on vehicle flow monitoring when running.

[0111] In the present application embodiment, Figure 2 As shown, the tunnel lighting intelligent control system 100 includes at least one processor 101, and at least one memory 102 and a bus 103 connected to the processor 101; wherein the processor 101 and the memory 102 communicate with each other through the bus 103; the processor 101 is used to call the program instructions in the memory 102 to execute the above-mentioned tunnel lighting intelligent control method based on vehicle flow monitoring.

[0112] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, tunnel lighting intelligent control systems (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] In a typical configuration, the tunnel lighting intelligent control system includes one or more processors (CPU), memory and bus. The tunnel lighting intelligent control system may also include input / output interface, network interface and the like.

[0114] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0115] Computer readable media include permanent and non-permanent, removable and non-removable media that can implement information storage by any method or technology. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage computer readable storage media or any other non-transmission media that can be used to store information that can be accessed by the tunnel lighting intelligent control system. As defined in this article, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carriers.

[0116] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or computer-readable storage medium that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such process, method, commodity or computer-readable storage medium. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the presence of other identical elements in the process, method, commodity or computer-readable storage medium that includes the elements.

[0117] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the embodiments of the present application can be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0118] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A tunnel lighting intelligent control method based on vehicle flow monitoring, characterized in that: The method is applied to a tunnel lighting intelligent control system, and the method comprises: Acquire the three-dimensional spatial information of the target tunnel to be matched with the lighting control strategy, the vehicle flow sensor monitoring information and the equipment network status information of the vehicle flow monitoring equipment network corresponding to the target tunnel; wherein the vehicle flow sensor monitoring information is the spatiotemporal domain monitoring information; the equipment network status information and the three-dimensional spatial information are the non-spatiotemporal domain monitoring information; According to the vehicle flow sensor monitoring information, determining the first correlation between the target tunnel and the plurality of initial lighting control strategies respectively through a debugged first control strategy generation network; Identify at least one tunnel segment space vector of the three-dimensional space information, and identify at least one device network state vector of the device network state information; wherein the tunnel segment space vector includes space dimension information, distribution density of lighting fixtures in different areas, and factors affecting light propagation by structures in different areas; the device network state vector includes sensor accuracy and data transmission stability; The lighting intelligent control strategy vector of the fusion result of each lighting control decision vector is determined by the debugged second control strategy generation network, and each lighting control decision vector is fully connected, including: determining the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector according to the first network weight of the debugged second control strategy generation network; determining the lighting intelligent control strategy weighted vector between the first network weight and the mapping characteristic value of each lighting control decision vector; determining the global decision vector of any two lighting control decision vectors between each lighting control decision vector according to the comparison result between the lighting intelligent control strategy vector of the fusion result and the determined lighting intelligent control strategy weighted vector; wherein, the mapping characteristic value of the lighting control decision vector is calculated based on the association between the characteristics of the lighting control decision vector and related factors, and the mapping characteristic value of the lighting control decision vector in the middle section is related to the expected traffic flow, lamp layout density and structure of the middle section; the method for determining the lighting intelligent control strategy weighted vector includes: multiplying and accumulating the corresponding weight value in the first network weight with the mapping characteristic value; Determine the second correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively; wherein the lighting control decision vector includes the at least one tunnel section space vector and the at least one device network state vector; the lighting intelligent control strategy vector of the fusion result is a vector that comprehensively reflects the relationship between the various lighting control decision vectors and the role of the various lighting control decision vectors in the overall lighting control; The target lighting control strategy of the target tunnel is determined by using the first correlation degree that the target tunnel is respectively matched with a plurality of initial lighting control strategies and the second correlation degree that the target tunnel is respectively matched with a plurality of the initial lighting control strategies.

2. The method according to claim 1, characterized in that The determining the target lighting control strategy of the target tunnel by using the first correlation degree that the target tunnel is respectively matched with a plurality of initial lighting control strategies and the second correlation degree that the target tunnel is respectively matched with a plurality of the initial lighting control strategies comprises: Determining a third correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively according to a first bias term coefficient of the first control strategy generating network and a second bias term coefficient of the second control strategy generating network, a first correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively, and a second correlation degree that the target tunnel matches the multiple initial lighting control strategies respectively; The initial lighting control strategy corresponding to the largest third correlation is used as the target lighting control strategy for the target tunnel.

3. The method according to claim 1, characterized in that The first control strategy generation network and the second control strategy generation network are debugged by the following steps: Acquire a training information set; the training information set includes three-dimensional space information samples corresponding to a plurality of tunnel samples, vehicle flow sensor monitoring information samples of a vehicle flow monitoring device network sample corresponding to each tunnel sample, and device network status information samples; Each tunnel sample has a priori lighting control strategy; Acquire a debugging example set and a verification example set from the training information set; wherein the debugging example set and the verification example set do not contain data corresponding to the same tunnel samples; Perform at least one debugging process on the original control strategy generation network according to the debugging example set, and verify the original control strategy generation network obtained by each debugging process according to the verification example set, and obtain the verification result corresponding to each debugging process, until the set debugging termination requirements are met, and determine the debugged first control strategy generation network and the second control strategy generation network based on the verification result corresponding to each debugging process; The original control strategy generation network includes a first original control strategy generation network corresponding to the first control strategy generation network and a second original control strategy generation network corresponding to the second control strategy generation network.

4. The method according to claim 3, characterized in that The debugging process includes: arbitrarily selecting a portion of the first historical examples from the debugging example set as the historical examples corresponding to the current debugging process; According to the vehicle flow sensor monitoring information sample in the first historical example, a first original control strategy generation network is used to determine that the tunnel sample matches the first correlation debugging values ​​of the multiple initial lighting control strategies respectively; Identify a lighting control decision vector sample for a second original control strategy generation network in the first historical example; the lighting control decision vector sample includes a tunnel sample segment space vector of a plurality of tunnel samples and a sample device network state vector corresponding to the tunnel sample; Based on the lighting control decision vector sample, determining through a second original control strategy generation network that the tunnel sample matches the second correlation debugging values ​​of the plurality of initial lighting control strategies respectively; Determining a lighting control strategy debugging result of the tunnel sample according to first correlation debugging values ​​of the tunnel sample that are respectively matched to a plurality of initial lighting control strategies, and second correlation debugging values ​​of the target tunnel that are respectively matched to a plurality of the initial lighting control strategies; According to the difference between the lighting control strategy debugging result and the prior lighting control strategy, the network weights of the original control strategy generation network are optimized, and the original control strategy generation network after the optimized network weights is used as the original control strategy generation network before the next debugging process.

5. The method according to claim 4, characterized in that The method of determining, based on the lighting control decision vector sample, that the tunnel sample matches the second correlation debugging values ​​of the multiple initial lighting control strategies respectively through the second original control strategy generation network comprises: Through the second original control strategy generation network, the lighting intelligent control strategy vector of the fusion result of each lighting control decision vector sample is determined, and the lighting intelligent control strategy vector of each lighting control decision vector sample is determined respectively. According to the lighting intelligent control strategy vector of the fusion result and the lighting intelligent control strategy vector of each lighting control decision vector, each lighting control decision vector sample is fully connected to determine the second correlation debugging value that the tunnel sample matches each of the multiple initial lighting control strategies.

6. The method according to claim 4, characterized in that The method of determining, based on the vehicle flow sensor monitoring information sample in the first historical example, generating a network through a first original control strategy, that the tunnel sample matches the first correlation debugging values ​​of the multiple initial lighting control strategies respectively includes: Determine a monitoring thermal intensity mark value in the vehicle flow sensor monitoring information sample that is not less than a first set monitoring thermal intensity; wherein the first set monitoring thermal intensity is determined based on a set vehicle flow coefficient in the vehicle flow sensor monitoring information sample; The monitoring thermal mark value in the vehicle flow sensor monitoring information sample is cleaned to obtain a first vehicle flow sensor monitoring information sample, and a lighting control decision vector for the first original control strategy generation network is determined based on the first vehicle flow sensor monitoring information sample.

7. The method according to claim 6, characterized in that The step of determining the lighting control decision vector for the first original control strategy generation network according to the first vehicle flow sensor monitoring information sample includes: Determine a second vehicle flow sensor monitoring information sample in which the sensor monitoring thermal value in the first vehicle flow sensor monitoring information sample is not less than a second set monitoring thermal intensity; wherein the second set monitoring thermal intensity is less than the first set monitoring thermal intensity; Updating the second vehicle flow sensor monitoring information sample to obtain a vehicle flow sensor monitoring update sample; In combination with the vehicle flow sensor monitoring update sample and the third vehicle flow sensor monitoring information sample, the lighting control decision vector of the first original control strategy generation network is determined; wherein the third vehicle flow sensor monitoring information sample is the vehicle flow sensor monitoring information sample in the first vehicle flow sensor monitoring information sample excluding the second vehicle flow sensor monitoring information sample.

8. An intelligent control system for tunnel lighting, characterized in that: It comprises a processor, a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call the program instructions in the memory to execute the intelligent control method for tunnel lighting based on vehicle flow monitoring according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by the processor, the intelligent control method for tunnel lighting based on vehicle flow monitoring as described in any one of claims 1 to 7 is implemented.

Images