Tunnel illumination intelligent control method and system

Abstract

The invention relates to the technical field of lighting control, in particular to an intelligent control method and system for tunnel lighting, and the method comprises the following steps: collecting traffic flow data, vehicle speed data and brightness outside a tunnel of a target tunnel; the method comprises the following steps: establishing a traffic flow prediction data set by using traffic flow data, establishing a vehicle speed prediction data set by using vehicle speed data, further constructing a traffic flow prediction model and a vehicle speed prediction model to obtain a traffic flow prediction value and a vehicle speed prediction value so as to formulate a tunnel illumination plan, and optimizing the tunnel illumination plan based on an image visibility strategy. And an optimal tunnel illumination scheme is obtained to illuminate the target tunnel. According to the method, the randomness and non-uniformity of tunnel illumination control influence factors are fully considered on the basis of introducing a machine learning algorithm, and the potential influence of light on the vision of a driver can be fully considered by using an image visibility strategy, so that the effect of tunnel illumination control is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of lighting control, and in particular to a tunnel lighting intelligent control method and system. Background Art

[0002] With the continuous development of transportation infrastructure, tunnels, as important transportation facilities, are receiving increasing attention for the performance and efficiency of their lighting systems. Tunnel lighting is not only related to driving safety, but also directly affects energy consumption and operating costs. Therefore, it is very important to effectively control the tunnel lighting system.

[0003] Traditional tunnel lighting control systems are mostly based on programmable logic controllers, which are relatively simple in control mode and usually perform hierarchical control according to the designed maximum traffic volume. This control mode is not intelligent enough and cannot make timely adjustments based on the actual situation of the tunnel environment, resulting in energy waste or poor lighting effects. At the same time, due to the particularity of tunnel lighting, drivers are prone to "black hole" and "white hole" effects when entering and exiting the tunnel, affecting driving safety.

[0004] In order to improve the intelligence and effect of tunnel lighting control, some experts have explored and practiced some tunnel lighting control schemes based on machine learning algorithms and fuzzy theory, and have achieved considerable results. However, the tunnel environment is complex and changeable, and factors such as traffic flow, vehicle speed, and light intensity outside the tunnel are highly random and uneven, and the existing tunnel lighting control methods have not yet considered them in more detail, resulting in unsatisfactory tunnel lighting control effects. In addition, the method of using light intensity outside the tunnel for lighting control also fails to fully consider the potential impact of light on the driver's vision, and cannot effectively alleviate the "black hole" and "white hole" effects, thereby further weakening the actual effect of tunnel lighting control. Summary of the invention

[0005] In view of the defects in the prior art, the present invention provides a tunnel lighting intelligent control method and system.

[0006] In order to achieve the above-mentioned purpose, in the first aspect, the present invention provides a method for intelligent control of tunnel lighting, the method comprising the following steps: setting a time interval for collecting data, and then collecting the traffic flow data and speed data of the target tunnel, and collecting the brightness outside the tunnel of the target tunnel; using the traffic flow data to establish a traffic flow prediction data set, and using the speed data to establish a speed prediction data set; constructing a traffic flow prediction model according to the traffic flow prediction data set to obtain a traffic flow prediction value, and constructing a speed prediction model according to the speed prediction data set to obtain a speed prediction value; formulating a tunnel lighting plan based on the traffic flow prediction value, the speed prediction value and the brightness outside the tunnel, and optimizing the tunnel lighting plan based on the image visibility strategy to obtain a tunnel lighting plan; illuminating the target tunnel according to the optimal tunnel lighting plan. The present invention fully considers the randomness and unevenness of the factors affecting tunnel lighting control on the basis of introducing a machine learning algorithm, and the image visibility strategy used can also fully take into account the potential impact of light on the driver's vision, thereby improving the effect of tunnel lighting control.

[0007] Optionally, the vehicle flow prediction data set includes a plurality of first vehicle flow prediction data sets and a plurality of second vehicle flow prediction data sets, and the vehicle speed prediction data set includes a plurality of first vehicle speed prediction data sets and a plurality of second vehicle speed prediction data sets; The step of using the vehicle flow data to establish a vehicle flow prediction data set and simultaneously using the vehicle speed data to establish a vehicle speed prediction data set comprises the following steps: The vehicle flow data is divided into vehicle flow data on weekdays and vehicle flow data on weekends, and the vehicle speed data is correspondingly divided into vehicle speed data on weekdays and vehicle speed data on weekends; Constructing a plurality of the first traffic flow prediction data sets according to the working day traffic flow data, and constructing a plurality of the second traffic flow prediction data sets according to the rest day traffic flow data; A plurality of the first vehicle speed prediction data sets are constructed according to the vehicle speed data on working days, and a plurality of the second vehicle speed prediction data sets are constructed according to the vehicle speed data on rest days.

[0008] Furthermore, dividing the traffic flow and speed data into two categories, weekdays and weekends, can help capture the periodic changes in traffic flow more accurately, alleviate the randomness and unevenness of traffic flow and speed distribution, and thus improve the accuracy of lighting control.

[0009] Optionally, constructing a plurality of the first traffic flow prediction data sets according to the weekday traffic flow data, and constructing a plurality of the second traffic flow prediction data sets according to the weekend traffic flow data comprises the following steps: Analyze the weekday traffic flow data, and then divide the weekday traffic flow into multiple weekday data change intervals from a time dimension; Using the weekday traffic flow data within the weekday data change interval to construct the first traffic flow prediction data set, and then obtain a plurality of the first traffic flow prediction data sets; Analyze the rest day traffic flow data, and then divide the rest day traffic flow into a plurality of rest day data change intervals from a time dimension; The second vehicle flow prediction data set is constructed using the holiday traffic flow data within the holiday data variation interval, thereby obtaining a plurality of the second vehicle flow prediction data sets.

[0010] Furthermore, by dividing the data change intervals by the time dimension, it is possible to more accurately capture the periodic changes in traffic flow, provide more reliable data support for model construction, and help lighting control to more flexibly adapt to changes in traffic flow and reduce the problem of insufficient or excessive lighting caused by uneven traffic flow distribution.

[0011] Optionally, constructing a plurality of the first vehicle speed prediction data sets according to the vehicle speed data on weekdays, and constructing a plurality of the second vehicle speed prediction data sets according to the vehicle speed data on weekends comprises the following steps: constructing the first vehicle speed prediction data set using the weekday vehicle speed data within the weekday data change interval, and thereby obtaining a plurality of the first vehicle speed prediction data sets; The second vehicle speed prediction data set is constructed using the rest day vehicle speed data within the rest day data variation interval, thereby obtaining a plurality of the second vehicle speed prediction data sets.

[0012] Furthermore, by dividing the data change intervals by the time dimension, the periodic changes in vehicle speed can be captured more accurately, providing more reliable data support for model construction, which is conducive to making lighting control more flexibly adapt to changes in vehicle speed and reducing insufficient or excessive lighting problems caused by uneven speed distribution.

[0013] Optionally, constructing a vehicle flow prediction model according to the vehicle flow prediction data set to obtain a vehicle flow prediction value, and constructing a vehicle speed prediction model according to the vehicle speed prediction data set to obtain a vehicle speed prediction value comprises the following steps: Improve the long short-term memory neural network to obtain the data prediction network; Using each of the first vehicle flow prediction data sets and each of the second vehicle flow prediction data sets to train and verify the data prediction network, respectively, to obtain the corresponding vehicle flow prediction model; Using each of the vehicle flow prediction models to obtain the vehicle flow prediction value within the corresponding working day data change interval and the rest day data change interval; Using each of the first vehicle speed prediction data sets and each of the second vehicle speed prediction data sets to train and verify the data prediction network, respectively, to obtain the corresponding vehicle speed prediction model; The vehicle speed prediction models are used to obtain the corresponding vehicle speed prediction values ​​within the working day data change interval and the rest day data change interval.

[0014] Furthermore, the present invention uses a long short-term memory neural network to construct prediction models for different time periods to predict vehicle flow and speed, so that lighting control can more flexibly adapt to changes in traffic flow, improve the intelligence level of lighting control, reduce insufficient or excessive lighting problems caused by uneven traffic flow, and improve the accuracy and reliability of lighting control.

[0015] Optionally, the improving the long short-term memory neural network to obtain a data prediction network comprises the following steps: A data trend enhancement module is introduced at the input end of the long short-term memory neural network to obtain a primary data prediction network. The function of the data trend enhancement module can be expressed using the following relationship: , Wherein, Y is the data sequence output by the data trend enhancement module, is the jth data in the data sequence output by the data trend enhancement module, , n is the length of the data sequence input into the data trend enhancement module, The i-th data in the data sequence input into the data trend enhancement module; The hyperparameters of the data prediction network are optimized using a sparrow search algorithm, thereby obtaining a final data prediction network, namely the data prediction network.

[0016] Furthermore, the present invention improves the performance of the long short-term memory neural network by improving the long short-term memory neural network, improving the accuracy of vehicle speed and traffic flow prediction, and thus improving the accuracy of lighting control.

[0017] Optionally, formulating a tunnel lighting plan according to the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, and optimizing the tunnel lighting plan based on an image visibility strategy to obtain a tunnel lighting plan includes the following steps: According to the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, preset values ​​of lighting brightness of each section of the target tunnel are set to obtain a tunnel lighting plan and illuminate the target tunnel; Two identical circular simple images are symmetrically arranged on the inner and outer walls of the target tunnel entrance, and two cameras are symmetrically arranged on both sides of the outer road of the target tunnel entrance; Using a camera to obtain an image of the circular simple picture on the opposite side and preprocessing it, thereby obtaining two grayscale images of the same size, which are recorded as a first grayscale image and a second grayscale image; A control group is set outside the target tunnel according to the circular simple picture and the arrangement of the camera at the entrance of the target tunnel to obtain a third grayscale image and a fourth grayscale image; According to an image fusion function, the first grayscale image and the second grayscale image are fused to obtain a first fused image, and the third grayscale image and the fourth grayscale image are fused to obtain a second fused image; Using an image visibility evaluation model, respectively calculating the image visibility of the first fused image and the second fused image, and recording them as first visibility and second visibility respectively; Recording a ratio of the first visibility to the second visibility as visual similarity, and setting an acceptable range of the visual similarity; If the real-time visual similarity is within the acceptable range, the tunnel lighting plan is used as the tunnel lighting solution; If the real-time visual similarity is not within the acceptable range, the preset lighting brightness value is adjusted until the real-time visual similarity falls within the acceptable range, thereby obtaining the tunnel lighting solution.

[0018] Furthermore, the present invention reflects the visual effects inside and outside the tunnel entrance by calculating the first visibility of the first fused image and the second visibility of the second fused image, and then calculates the visual similarity to optimize the tunnel lighting plan, and then considers the potential impact of the light inside and outside the tunnel entrance on the driver's vision when performing lighting control, and can respond quickly when the natural light intensity changes rapidly, effectively alleviating the "black hole" and "white hole" effects, and improving driving safety and comfort.

[0019] Optionally, the image fusion function satisfies the following relationship: , in, is the mth column of pixels from left to right on the image, is the kth row and the The grayscale value of the pixel in the column, is the mth column of pixels from right to left on the image, is the first image in the kth row. The grayscale value of the pixel in the column, is the kth row and the first row of the second image involved in the fusion The grayscale value of the pixel in the column.

[0020] Optionally, the image visibility evaluation model satisfies the following relationship: , in, is the image visibility, is the average gray value of the image, is the number of background pixels on the image, is the gray value of the dth background pixel on the image, is the average gray value of the pattern on the image.

[0021] Furthermore, the image visibility evaluation model comprehensively considers factors such as the average gray value of the image, the number of background pixels and the average gray value of the pattern, and can objectively evaluate the visibility of the pattern, thereby improving the accuracy of lighting control.

[0022] In a second aspect, the present invention provides a tunnel lighting intelligent control system, the tunnel lighting intelligent control system comprising: a data acquisition device, a data output device, a processor and a storage, the storage comprising a computer-readable storage medium, the computer-readable storage medium storing a computer program, the computer program comprising program instructions, the program instructions when executed by the processor enable the processor to implement the tunnel lighting intelligent control method of the present invention. This system is adapted to the method provided by the present invention, and it not only has the advantages of the method provided by the present invention, but also can improve the practicality of the method provided by the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] 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 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 paying creative work.

[0024] Figure 1 A schematic diagram showing a flow chart of an intelligent control method for tunnel lighting according to an embodiment of the present invention; Figure 2 The figure is a schematic diagram of the framework of a tunnel lighting intelligent control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that these specific details do not need to be adopted to implement the present invention. In other examples, in order to avoid confusing the present invention, known circuits, software or methods are not specifically described.

[0026] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or subcombination. In addition, it should be understood by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale.

[0027] It should be noted in advance that, in an optional embodiment, except for independent explanations, the same symbols or letters appearing in all formulas have the same meanings and values.

[0028] In an alternative embodiment, see Figure 1 The present invention provides a tunnel lighting intelligent control method, the method comprising the following steps: S1. Setting a time interval for data collection, and then collecting the traffic volume data and vehicle speed data of the target tunnel, and at the same time collecting the brightness outside the target tunnel.

[0029] Specifically, in this embodiment, the time interval is 5 minutes, the traffic flow data refers to the total number of vehicles passing through the target tunnel in each time interval, the speed data refers to the maximum speed of all vehicles at the entrance of the target tunnel in each time interval, and the brightness outside the tunnel refers to the natural light brightness outside the entrance of the target tunnel.

[0030] Furthermore, the traffic volume data can be collected by installing a vehicle detector at the entrance of the target tunnel, the vehicle speed data can be collected by installing a radar speed meter at the entrance of the target tunnel, and the brightness outside the tunnel can be collected by installing a light sensor near the entrance of the target tunnel. The models of the vehicle detector, radar speed meter and light sensor can be selected according to actual conditions. In other optional embodiments, the traffic volume data, vehicle speed data and brightness outside the tunnel can also be collected by other means.

[0031] S2. Using the vehicle flow data to establish a vehicle flow prediction data set, and using the vehicle speed data to establish a vehicle speed prediction data set.

[0032] The vehicle flow prediction data set includes a plurality of first vehicle flow prediction data sets and a plurality of second vehicle flow prediction data sets, and the vehicle speed prediction data set includes a plurality of first vehicle speed prediction data sets and a plurality of second vehicle speed prediction data sets. Step S2 specifically includes the following steps: S21. Divide the vehicle flow data into vehicle flow data on working days and vehicle flow data on rest days, and correspondingly divide the vehicle speed data into vehicle speed data on working days and vehicle speed data on rest days.

[0033] Specifically, in this embodiment, the traffic volume and speed in the same tunnel usually show strong randomness and unevenness. Considering that there is a significant difference in tunnel traffic volume on weekdays and weekends, this embodiment divides the traffic volume data and speed data into two categories: weekdays and weekends. This helps to more accurately capture the periodic changes in traffic flow, alleviate the randomness and unevenness of traffic flow and speed distribution, and thus improve the accuracy of lighting control.

[0034] Furthermore, there is a strong coupling between traffic volume and vehicle speed. Generally speaking, the greater the traffic volume, the more likely the driver will reduce the speed in order to ensure driving safety, and vice versa. Therefore, the vehicle speed data is divided into weekday speed data and weekend speed data.

[0035] S22: construct a plurality of the first traffic flow prediction data sets according to the weekday traffic flow data, and construct a plurality of the second traffic flow prediction data sets according to the weekend traffic flow data.

[0036] This step divides the data change interval from the time dimension and establishes multiple first traffic flow prediction data sets and second traffic flow prediction data sets, which can more accurately capture the periodic changes in traffic flow and provide more reliable data support for the subsequent construction of a traffic flow prediction model, which is conducive to making lighting control more flexible to adapt to changes in traffic flow and reduce the problem of insufficient or excessive lighting caused by uneven distribution of traffic flow. Step S22 specifically includes the following steps: S221. Analyze the weekday traffic flow data, and then divide the weekday traffic flow into a plurality of weekday data change intervals from a time dimension.

[0037] Specifically, in this embodiment, the weekday traffic flow data of the target tunnel is analyzed, and then the weekday traffic flow is divided into four weekday data change intervals from the time dimension, namely the weekday low peak period, the weekday morning peak period, the weekday flat peak period and the weekday evening peak period. Among them, the weekday low peak period occurs between 21:00 and 6:00, when most people are still resting and the traffic flow is small; the weekday morning peak period occurs between 6:00 and 10:00, when people start to get up and go to work, study, etc., and the traffic flow in the target tunnel will increase rapidly and reach the first peak of the day, that is, the morning peak; the weekday flat peak period occurs between 10:00 and 17:00, when most people are working or studying, so the traffic flow in the target tunnel is relatively stable; the weekday evening peak period occurs between 17:00 and 21:00, which is the time period when people return in large numbers. At this time, the traffic flow in the target tunnel reaches the second peak of the day, that is, the evening peak, and sometimes the evening peak can even exceed the morning peak. It should be noted that the specific division of the weekday data change interval may vary depending on the region and the target tunnel.

[0038] S222: Use the weekday traffic flow data within the weekday data change interval to construct the first traffic flow prediction data set, and then obtain multiple first traffic flow prediction data sets.

[0039] Specifically, in this embodiment, by continuously collecting the traffic flow data of the target tunnel in real time, a large amount of workday traffic flow data of different workdays can be obtained, so the workday traffic flow data within the same workday data change interval in each workday can be used to construct a first traffic flow prediction data set. Since step S221 divides the workday data change interval into four, four first traffic flow prediction data sets can be constructed.

[0040] S223: Analyze the rest day traffic flow data, and further divide the rest day traffic flow into a plurality of rest day data change intervals from a time dimension.

[0041] Specifically, in this embodiment, compared with weekdays, people do not have fixed commuting times on weekends, the low and high peak periods of traffic flow are not very obvious, the traffic flow is more evenly distributed throughout the day than on weekdays, and the change in traffic flow is also slightly more gentle. Therefore, this embodiment divides the traffic flow on weekends into two data change intervals of daytime and nighttime. The daytime refers to 6:00 to 20:00, and the nighttime refers to 20:00 to 6:00. Similarly, the division of the specific data change intervals of weekends may vary depending on the region and season.

[0042] S224: Use the rest day traffic flow data within the rest day data variation interval to construct the second traffic flow prediction data set, and then obtain a plurality of the second traffic flow prediction data sets.

[0043] Specifically, in this embodiment, since step S223 divides the holiday traffic flow into two holiday data change intervals, namely, daytime and nighttime, two second traffic flow prediction data sets can be established.

[0044] S23: construct a plurality of first vehicle speed prediction data sets according to the weekday vehicle speed data, and construct a plurality of second vehicle speed prediction data sets according to the weekend vehicle speed data.

[0045] This step divides the data change interval from the time dimension and establishes multiple first vehicle speed prediction data sets and second vehicle speed prediction data sets, which can more accurately capture the periodic changes in vehicle speed and provide more reliable data support for the subsequent construction of a vehicle speed prediction model, which is conducive to making the lighting control more flexible to adapt to changes in vehicle speed and reduce the problem of insufficient or excessive lighting caused by uneven speed distribution. Step S23 specifically includes the following steps: S231. Use the weekday vehicle speed data within the weekday data change interval to construct the first vehicle speed prediction data set, and then obtain a plurality of the first vehicle speed prediction data sets.

[0046] Specifically, in this embodiment, according to the four working day data change intervals divided in step S221, four corresponding first vehicle speed prediction data sets can be established.

[0047] S232: construct the second vehicle speed prediction data set using the rest day vehicle speed data within the rest day data variation interval, and further obtain a plurality of the second vehicle speed prediction data sets.

[0048] Specifically, in this embodiment, two corresponding second vehicle speed prediction data sets may be established according to the two rest day data change intervals divided in step S223.

[0049] S3. Constructing a vehicle flow prediction model according to the vehicle flow prediction data set to obtain a vehicle flow prediction value, and constructing a vehicle speed prediction model according to the vehicle speed prediction data set to obtain a vehicle speed prediction value.

[0050] Wherein, step S3 specifically includes the following steps: S31. Improve the long short-term memory neural network to obtain a data prediction network.

[0051] Wherein, step S31 specifically includes the following steps: S311. Introduce a data trend enhancement module at the input end of the long short-term memory neural network to obtain a primary data prediction network.

[0052] The function of the data trend enhancement module can be expressed using the following relationship: , Among them, Y is the data sequence output by the data trend improvement module, is the jth data in the data sequence output by the data trend enhancement module. , n is the length of the data sequence of the input data trend enhancement module, It is the i-th data in the data sequence of the input data trend enhancement module.

[0053] Furthermore, although step S2 can alleviate the problem of insufficient or excessive lighting caused by the strong randomness and unevenness of traffic flow and speed to a certain extent, in a short period of time, especially during peak hours, traffic flow and speed may still fluctuate with a high frequency, that is, traffic flow and speed have strong randomness. Therefore, this embodiment uses a data trend enhancement module to process the traffic flow data sequence or speed data sequence input into the long short-term memory neural network to improve the trend of the data sequence, and also helps to improve the reliability of the output results of the traffic flow prediction model and the speed prediction model.

[0054] S312: Use a sparrow search algorithm to optimize the hyperparameters of the data prediction network, and then obtain a final data prediction network, that is, the data prediction network.

[0055] Specifically, in this embodiment, the hyperparameters of the data prediction network, i.e., the hyperparameters of the long short-term memory neural network, include the number of hidden layer nodes, the number of iterations, the initial learning rate, the learning rate reduction factor, and the reduction segment length. Using the sparrow search algorithm to optimize the hyperparameters of the data prediction network is a prior art and will not be described in detail here.

[0056] S32: Use each of the first vehicle flow prediction data sets and each of the second vehicle flow prediction data sets to train and verify the data prediction network to obtain the corresponding vehicle flow prediction model.

[0057] Specifically, in this embodiment, each first traffic flow prediction data set and each second traffic flow prediction data set are used to train and verify the data prediction network, which can eventually be used to predict the traffic flow prediction model in each weekday data change interval or rest day data change interval.

[0058] More specifically, for example, the data prediction network is trained and verified using the first traffic flow prediction data set during the off-peak period on weekdays, and ultimately a traffic flow prediction model can be obtained, which can then be used to predict the traffic flow prediction value during the off-peak period on weekdays.

[0059] S33: Utilize each of the vehicle flow prediction models to obtain the vehicle flow prediction values ​​within the corresponding working day data change interval and the rest day data change interval.

[0060] Specifically, in this embodiment, taking a certain working day data change interval as an example, since the time interval used in step S1 is 5 minutes, when the traffic flow prediction model corresponding to the working day data change interval is used to predict the traffic flow, the prediction result is actually the traffic flow prediction value for the next 5 minutes within the working day data change interval, and the prediction starts from the last 30 seconds of the current 5 minutes.

[0061] S34. Use each of the first vehicle speed prediction data sets and each of the second vehicle speed prediction data sets to train and verify the data prediction network to obtain the corresponding vehicle speed prediction model.

[0062] Specifically, in this embodiment, this step is similar to step S32 and will not be described in detail here.

[0063] S35. Utilize each of the vehicle speed prediction models to obtain the corresponding vehicle speed prediction values ​​within the working day data change interval and the rest day data change interval.

[0064] Specifically, in this embodiment, taking a certain working day data change interval as an example, when using the vehicle speed prediction model corresponding to the working day data change interval to predict traffic flow, the prediction result is actually the vehicle speed prediction value for the next 5 minutes within the working day data change interval, and the prediction starts from the last 30 seconds of the current 5 minutes.

[0065] S4. Formulate a tunnel lighting plan based on the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, and optimize the tunnel lighting plan based on an image visibility strategy to obtain a tunnel lighting solution.

[0066] Wherein, step S4 specifically includes the following steps: S41, setting preset values ​​of lighting brightness of each section of the target tunnel according to the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, obtaining a tunnel lighting plan and illuminating the target tunnel.

[0067] Specifically, in this embodiment, the target tunnel is divided into an entrance section, a first transition section, a middle section, a second transition section, and an exit section, and the vehicle passes through the entrance section, the first transition section, the middle section, the second transition section, and the exit section in sequence. Among them, the preset values ​​of the lighting brightness of the entrance section and the exit section are the same, and the preset values ​​of the lighting brightness of the first transition section and the second transition section are the same. The preset values ​​of the lighting brightness of the entrance section, the first transition section, and the middle section under different vehicle flow prediction values, vehicle speed prediction values, and tunnel brightness respectively satisfy the following relationship:

[0068]

[0069]

[0070] in, , and are the preset values ​​of the lighting brightness of the entrance section, the first transition section and the middle section, respectively. L is the brightness outside the cave. is the predicted vehicle speed, is the historical maximum value of vehicle speed, M is the predicted value of vehicle flow, is the historical maximum value of traffic volume. Indicates the value and numerical values The larger value of Indicates the value and numerical values The smaller value of Maximum brightness designed for the target tunnel, is the minimum brightness in the middle segment, The value of can refer to the "Highway Tunnel Ventilation Lighting Design Specifications". The unit of brightness is " ", the unit of speed is "km / h" and the unit of traffic flow is "vehicles / 5 minutes".

[0071] Furthermore, strictly speaking, because the vehicle speed and traffic volume are predicted values, when setting the preset values ​​of the lighting brightness of each section of the target tunnel, the brightness outside the tunnel should also use the predicted values. However, considering that the change of the brightness outside the tunnel is very small in a short period of time, the brightness outside the tunnel used when setting the preset values ​​of the lighting brightness of each section of the target tunnel is the brightness outside the tunnel measured in real time. In addition, the use of the brightness outside the tunnel measured in real time can also solve the negative impact of the randomness of the change of the light intensity outside the target tunnel on the accuracy of the lighting control in the tunnel to a certain extent.

[0072] S42, symmetrically setting two identical circular simple images on the inner side walls of the target tunnel entrance, and symmetrically setting two cameras on both sides of the outer road of the target tunnel entrance.

[0073] Specifically, in this embodiment, two identical circular simple images are symmetrically arranged on the inner wall surfaces on both sides of the target tunnel entrance, and are respectively recorded as the first image and the second image. The diameter of the circular simple images is 0.4 meters, and the distance between them and the target tunnel entrance is 6 meters. Two cameras are symmetrically arranged on both sides of the outer road of the target tunnel entrance, and are recorded as the first camera and the second camera. The distance between them and the target tunnel entrance is 5 meters. The first image and the first camera are located on the same side of the road, and the second image and the second camera are located on the same side of the road, and the distance between the first image and the second camera is the same as the distance between the second image and the first camera.

[0074] More specifically, the circular simple picture includes two parts: a background and a pattern. The background color of the circular simple picture is the same as the color of the tunnel wall, and the pattern on the circular simple picture is a red "Z"-shaped pattern. In addition, in order to facilitate the subsequent acquisition of the image of the circular simple picture, the circular simple picture can be placed in a photo frame and then symmetrically installed on the inner walls of the target tunnel entrance.

[0075] Furthermore, in other optional embodiments, the shape, background and pattern of the pictures symmetrically arranged on the inner walls on both sides of the tunnel entrance can be designed by oneself, and the distance between the pictures and the target tunnel entrance and the distance between the camera and the target tunnel entrance can be selected according to the actual situation of the target tunnel. In addition, if the target tunnel does not meet the conditions for symmetrically arranging pictures on the inner walls on both sides of the entrance or the conditions for symmetrically arranging cameras on both sides of the outer road of the entrance do not meet the conditions for symmetrically arranging cameras, pictures can be arranged only on one side of the inner wall of the target tunnel entrance, and a camera can be arranged on one side of the outer road of the target tunnel entrance, and the arranged pictures and camera are not on the same side of the road.

[0076] S43, using a camera to obtain an image of the circular simple picture on the opposite side and perform preprocessing, thereby obtaining two grayscale images of the same size, which are recorded as a first grayscale image and a second grayscale image.

[0077] Specifically, in this embodiment, the first camera is used to capture the second picture in real time to obtain the image of the second picture. After obtaining the image of the second picture, the image needs to be preprocessed, specifically including image deblurring, grayscale and image cropping, and finally obtains the first grayscale image. Among them, image cropping is to crop the grayscale image to a specific size.

[0078] Furthermore, the first picture is captured in real time using a second camera, an image of the first picture is obtained and preprocessed to obtain a second grayscale image.

[0079] S44, setting a control group outside the target tunnel according to the circular simple picture and the arrangement of the cameras at the entrance of the target tunnel to obtain a third grayscale image and a fourth grayscale image.

[0080] Specifically, in this embodiment, the third picture, the fourth picture, the third camera and the fourth camera are arranged outside the target tunnel as a control group according to the arrangement of the first picture, the second picture, the first camera and the second camera, and then the third grayscale image and the fourth grayscale image are obtained in the same manner as the first grayscale image and the second grayscale image. Among them, the third picture and the third camera are located on the same side of the road, the fourth picture and the fourth camera are located on the same side of the road, the third picture and the fourth picture are exactly the same as the first picture, the third camera and the fourth camera are exactly the same as the first camera, and the straight line where the third camera and the fourth camera are located is parallel to the straight line where the first camera and the second camera are located.

[0081] S45. According to an image fusion function, the first grayscale image and the second grayscale image are fused to obtain a first fused image, and the third grayscale image and the fourth grayscale image are fused to obtain a second fused image.

[0082] Specifically, in this embodiment, the image fusion function satisfies the following relationship:

[0083] in, is the mth column of pixels from left to right on the image, is the kth row and the The grayscale value of the pixel in the column, is the mth column of pixels from right to left on the image, is the first image in the kth row. The grayscale value of the pixel in the column, is the kth row and the first row of the second image involved in the fusion The grayscale value of the pixel in the column.

[0084] Furthermore, at the entrance of the target tunnel, affected by the angle of natural light, the first grayscale image and the second grayscale image obtained will be quite different. Therefore, it is difficult to accurately reflect the actual visual effect at the entrance of the tunnel through any grayscale image. Considering that the first grayscale image and the second grayscale image can be approximately mirror images of each other, this embodiment fuses the first grayscale image and the second grayscale image based on the image fusion function to obtain a first fused image, thereby improving this problem to a certain extent. At the same time, this embodiment fuses the third grayscale image and the fourth grayscale image to obtain a second fused image for comparison with the first fused image.

[0085] S46: Use an image visibility evaluation model to respectively calculate the image visibility of the first fused image and the second fused image, and record them as first visibility and second visibility respectively.

[0086] Specifically, in this embodiment, the image visibility evaluation model satisfies the following relationship:

[0087] in, For image visibility, is the average gray value of the image, is the number of background pixels on the image, is the gray value of the dth background pixel on the image, is the average gray value of the pattern on the image.

[0088] Furthermore, the optical camera can capture the real difference in light inside and outside the target tunnel and reflect this difference in the image it collects. In the image visibility evaluation model, Can reflect the brightness of the image surface, The larger the value, the easier it is for the driver to see various objects in the target tunnel when entering the target tunnel; It can reflect the uniformity of brightness around the pattern on a simple circular picture. The smaller the value, the better the uniformity of the brightness in the area around the pattern, which means that the driver can see the various objects in the target tunnel more clearly when entering the target tunnel. Therefore, the image visibility calculated using the image visibility evaluation model can more comprehensively reflect the impact of light on the driver's vision.

[0089] S47: Record the ratio of the first visibility to the second visibility as visual similarity, and set an acceptable range of the visual similarity.

[0090] Specifically, in this embodiment, by calculating the visual similarity, the similarity of the impact of the light inside and outside the target tunnel entrance on the driver's vision can be reflected. The closer the visual similarity is to 1, the more similar the light environment inside and outside the tunnel is to the driver, and the driver is less likely to experience a "black hole" effect when entering the target tunnel.

[0091] Considering the inevitable differences between the environments inside and outside the target tunnel, this embodiment sets the acceptable interval of visual similarity to [0.7, 1.1].

[0092] S48: If the real-time visual similarity is within the acceptable range, the tunnel lighting plan is used as the tunnel lighting solution.

[0093] S49: If the real-time visual similarity is not within the acceptable range, adjust the preset lighting brightness value until the real-time visual similarity falls within the acceptable range, thereby obtaining the tunnel lighting solution.

[0094] Specifically, in this embodiment, if the real-time visual similarity exceeds the upper limit of the acceptable interval, the preset value of the lighting brightness of the entrance section is continuously reduced until the real-time visual similarity falls into the acceptable interval, and then the preset values ​​of the lighting brightness of the first transition section and the middle section are adjusted according to the formula disclosed in step S41 based on the preset value of the lighting brightness of the entrance section after adjustment, and then the preset values ​​of the lighting brightness of the second transition section and the exit section are adjusted. If the real-time visual similarity is lower than the lower limit of the acceptable interval, the preset value of the lighting brightness of the entrance section is continuously increased until the real-time visual similarity falls into the acceptable interval or reaches the maximum brightness of the target tunnel design, and then the preset values ​​of the lighting brightness of the first transition section and the middle section are adjusted using the formula disclosed in step S41, and then the preset values ​​of the lighting brightness of the second transition section and the exit section are adjusted.

[0095] The preset value of the lighting brightness of each section after adjustment is recorded as the final value of the lighting brightness of each section. The final value of the lighting brightness of each section of the target tunnel is the tunnel lighting plan. Using this plan to illuminate the target tunnel can effectively alleviate the "black hole" and "white hole" effects, improve driving safety and comfort, and improve the effect of tunnel lighting control.

[0096] S5. Light the target tunnel according to the tunnel lighting scheme.

[0097] It should be noted that, in some cases, the actions described in the specification can be performed in a different order and still achieve the desired results. In this embodiment, the order of steps given is only to make the embodiment appear clearer and easier to explain, rather than to limit it.

[0098] In an alternative embodiment, see Figure 2 In order to improve the practicality of the method provided by the present invention, the present invention also provides a tunnel lighting intelligent control system, which includes: a data acquisition device 1, a data output device 2, a processor 3 and a storage 4, wherein the storage 4 includes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by the processor 3, the processor 3 implements the contents described in steps S1 to S5.

[0099] Specifically, in this embodiment, the data acquisition device 1, the data output device 2, the processor 3 and the storage 4 are electrically connected to each other. The data acquisition device 1 includes a vehicle detector, a radar speed meter, a light sensor and a camera. The data output device 2 is used to output the tunnel lighting solution.

[0100] In summary, this method divides the data change interval from the time dimension and establishes multiple first speed prediction data sets, second speed prediction data sets, first speed prediction data sets and second speed prediction data sets, which can more accurately capture the periodic changes of traffic flow and speed, and provide more reliable data support for the subsequent construction of traffic flow prediction model and speed prediction model, which is conducive to making lighting control more flexible to adapt to changes in traffic flow and speed, and reduce the problem of insufficient or excessive lighting caused by uneven distribution of traffic flow and speed. Further, based on the constructed data set, this method uses an improved long short-term memory neural network to build a prediction model to predict traffic flow and speed in different time periods, which improves the accuracy and reliability of traffic flow and speed prediction. Finally, based on the predicted traffic flow and speed, combined with the real-time collected outside tunnel brightness, this method obtains the tunnel lighting plan, and considers the potential impact of the light inside and outside the tunnel entrance on the driver's vision. Based on the image visibility strategy, the tunnel lighting plan is optimized, and then the tunnel is illuminated, effectively alleviating the "black hole" and "white hole" effects, improving driving safety and comfort, and improving the effect of tunnel lighting control.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A tunnel lighting intelligent control method, characterized in that: The steps include: Setting a time interval for data collection, thereby collecting traffic flow data and vehicle speed data of a target tunnel, and collecting the brightness outside the target tunnel at the same time; Using the vehicle flow data to establish a vehicle flow prediction data set, and using the vehicle speed data to establish a vehicle speed prediction data set; Constructing a vehicle flow prediction model according to the vehicle flow prediction data set to obtain a vehicle flow prediction value, and constructing a vehicle speed prediction model according to the vehicle speed prediction data set to obtain a vehicle speed prediction value; Formulate a tunnel lighting plan according to the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, and optimize the tunnel lighting plan based on an image visibility strategy to obtain a tunnel lighting solution; The target tunnel is illuminated according to the tunnel lighting plan.

2. The intelligent control method for tunnel lighting according to claim 1, characterized in that: The vehicle flow prediction data set includes a plurality of first vehicle flow prediction data sets and a plurality of second vehicle flow prediction data sets, and the vehicle speed prediction data set includes a plurality of first vehicle speed prediction data sets and a plurality of second vehicle speed prediction data sets; The step of using the vehicle flow data to establish a vehicle flow prediction data set and simultaneously using the vehicle speed data to establish a vehicle speed prediction data set comprises the following steps: The vehicle flow data is divided into vehicle flow data on weekdays and vehicle flow data on weekends, and the vehicle speed data is correspondingly divided into vehicle speed data on weekdays and vehicle speed data on weekends; Constructing a plurality of the first traffic flow prediction data sets according to the working day traffic flow data, and constructing a plurality of the second traffic flow prediction data sets according to the rest day traffic flow data; A plurality of the first vehicle speed prediction data sets are constructed according to the vehicle speed data on working days, and a plurality of the second vehicle speed prediction data sets are constructed according to the vehicle speed data on rest days.

3. The intelligent control method for tunnel lighting according to claim 2, characterized in that: The step of constructing a plurality of the first traffic flow prediction data sets according to the working day traffic flow data and constructing a plurality of the second traffic flow prediction data sets according to the rest day traffic flow data comprises the following steps: Analyze the weekday traffic flow data, and then divide the weekday traffic flow into multiple weekday data change intervals from a time dimension; Using the weekday traffic flow data within the weekday data change interval to construct the first traffic flow prediction data set, and then obtain a plurality of the first traffic flow prediction data sets; Analyze the rest day traffic flow data, and then divide the rest day traffic flow into a plurality of rest day data change intervals from a time dimension; The second vehicle flow prediction data set is constructed using the holiday traffic flow data within the holiday data variation interval, thereby obtaining a plurality of the second vehicle flow prediction data sets.

4. The intelligent control method for tunnel lighting according to claim 3, characterized in that: The step of constructing a plurality of the first vehicle speed prediction data sets according to the vehicle speed data on working days and constructing a plurality of the second vehicle speed prediction data sets according to the vehicle speed data on rest days comprises the following steps: constructing the first vehicle speed prediction data set using the weekday vehicle speed data within the weekday data change interval, and thereby obtaining a plurality of the first vehicle speed prediction data sets; The second vehicle speed prediction data set is constructed using the rest day vehicle speed data within the rest day data variation interval, thereby obtaining a plurality of the second vehicle speed prediction data sets.

5. The intelligent control method for tunnel lighting according to claim 4, characterized in that: The step of constructing a vehicle flow prediction model according to the vehicle flow prediction data set to obtain a vehicle flow prediction value, and constructing a vehicle speed prediction model according to the vehicle speed prediction data set to obtain a vehicle speed prediction value comprises the following steps: Improve the long short-term memory neural network to obtain the data prediction network; Using each of the first vehicle flow prediction data sets and each of the second vehicle flow prediction data sets to train and verify the data prediction network, respectively, to obtain the corresponding vehicle flow prediction model; Using each of the vehicle flow prediction models to obtain the vehicle flow prediction value within the corresponding working day data change interval and the rest day data change interval; Using each of the first vehicle speed prediction data sets and each of the second vehicle speed prediction data sets to train and verify the data prediction network, respectively, to obtain the corresponding vehicle speed prediction model; The vehicle speed prediction models are used to obtain the corresponding vehicle speed prediction values ​​within the working day data change interval and the rest day data change interval.

6. The intelligent control method for tunnel lighting according to claim 5, characterized in that: The improvement of the long short-term memory neural network to obtain a data prediction network includes the following steps: A data trend enhancement module is introduced at the input end of the long short-term memory neural network to obtain a primary data prediction network. The function of the data trend enhancement module can be expressed using the following relationship: ， Wherein, Y is the data sequence output by the data trend enhancement module, is the jth data in the data sequence output by the data trend enhancement module, , n is the length of the data sequence input into the data trend enhancement module, The i-th data in the data sequence input into the data trend enhancement module; The hyperparameters of the data prediction network are optimized using a sparrow search algorithm, thereby obtaining a final data prediction network, namely the data prediction network.

7. The intelligent control method for tunnel lighting according to claim 1, characterized in that: The step of formulating a tunnel lighting plan based on the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, and optimizing the tunnel lighting plan based on an image visibility strategy to obtain a tunnel lighting plan includes the following steps: According to the predicted value of vehicle flow, the predicted value of vehicle speed and the brightness outside the tunnel, preset values ​​of lighting brightness of each section of the target tunnel are set to obtain a tunnel lighting plan and illuminate the target tunnel; Two identical circular simple images are symmetrically arranged on the inner and outer walls of the target tunnel entrance, and two cameras are symmetrically arranged on both sides of the outer road of the target tunnel entrance; Use a camera to obtain an image of the circular simple picture on the opposite side and perform preprocessing, thereby obtaining two grayscale images of the same size, which are recorded as a first grayscale image and a second grayscale image; A control group is set outside the target tunnel according to the circular simple picture and the arrangement of the camera at the entrance of the target tunnel to obtain a third grayscale image and a fourth grayscale image; According to an image fusion function, the first grayscale image and the second grayscale image are fused to obtain a first fused image, and the third grayscale image and the fourth grayscale image are fused to obtain a second fused image; Using an image visibility evaluation model, respectively calculating the image visibility of the first fused image and the second fused image, and recording them as first visibility and second visibility respectively; Recording a ratio of the first visibility to the second visibility as visual similarity, and setting an acceptable range of the visual similarity; If the real-time visual similarity is within the acceptable range, the tunnel lighting plan is used as the tunnel lighting solution; If the real-time visual similarity is not within the acceptable range, the preset lighting brightness value is adjusted until the real-time visual similarity falls within the acceptable range, thereby obtaining the tunnel lighting solution.

8. The intelligent control method for tunnel lighting according to claim 7, characterized in that: The image fusion function satisfies the following relationship: ， in, is the mth column of pixels from left to right on the image, is the kth row and the The grayscale value of the pixel in the column, is the mth column of pixels from right to left on the image, is the first image in the kth row. The grayscale value of the pixel in the column, is the kth row and the first row of the second image involved in the fusion The grayscale value of the pixel in the column.

9. The intelligent control method for tunnel lighting according to claim 7, characterized in that: The image visibility evaluation model satisfies the following relationship: ， in, is the image visibility, is the average gray value of the image, is the number of background pixels on the image, is the gray value of the dth background pixel on the image, is the average gray value of the pattern on the image.

10. A tunnel lighting intelligent control system, characterized in that: The tunnel lighting intelligent control system comprises: a data acquisition device, a data output device, a processor and a storage device, wherein the storage device comprises a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program comprises program instructions, and when the program instructions are executed by the processor, the processor implements the tunnel lighting intelligent control method according to any one of claims 1 to 9.

Images