Tunnel time sequence dimming system integrating sensor data and data migration

Abstract

The invention discloses a tunnel time sequence dimming system integrating sensor data and data migration. The tunnel time sequence dimming system comprises a sensor module, a data storage module, a data processing and verification module, a dimming module, an illumination control module, a data monitoring and warning module and a data interface module. Therefore, a more perfect tunnel time sequence dimming system is constructed: 1, a migration strategy is innovatively introduced, an existing dimming model of a tunnel with similar information such as environment and traffic flow is used for migration training of a target tunnel, and a time sequence dimming model suitable for the target tunnel is quickly generated; 2, a data processing and verification module is added, intelligent processing of detection data is realized, the generalization ability of the model is improved, and stable operation of the system in a complex environment is ensured; and 3, a data monitoring and alarm module is added, the state of the sensor is monitored in real time, intelligent broadcasting of equipment abnormity is realized, maintenance personnel are reminded to overhaul or replace equipment in time, and the operation stability and the maintenance efficiency of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel lighting, and in particular to a tunnel sequential dimming system integrating sensor data and data migration. Background Art

[0002] With the rapid development of smart cities and intelligent transportation systems, tunnels, as an important part of transportation infrastructure, are becoming a key area for traffic safety and management, especially in mountainous and hilly areas. The lighting system in the tunnel is crucial to driving safety. Reasonable tunnel lighting can not only ensure the driver's clear vision, but also effectively reduce the occurrence of traffic accidents. However, most traditional tunnel lighting systems use a constant brightness control method, ignoring dynamic factors such as vehicle speed, traffic volume, and weather conditions, resulting in energy waste and increased system maintenance costs; at the same time, the traditional dimming method fails to perform intelligent dimming according to the actual on-site environment, and the large difference in light intensity inside and outside the tunnel is common, which can easily cause "white hole" and "black hole" effects to drivers, posing safety hazards; in addition, the lighting requirements of different areas in the tunnel are also different, and the lighting mode with fixed light intensity cannot provide drivers with a comfortable driving experience. For example, some places are too bright, causing extra glare, while some places are not bright, causing blurred vision.

[0003] In order to solve the above problems, a tunnel timing dimming system based on timing analysis and intelligent dimming technology came into being. The system collects sensor data inside and outside the tunnel in real time, including multi-dimensional information such as traffic flow, ambient light intensity, and weather conditions, and combines data analysis and intelligent prediction models to finely and dynamically adjust the lights in the tunnel. In different time periods and specific conditions, the system can automatically adjust the lighting brightness to achieve the purpose of energy saving and consumption reduction, while ensuring driving safety. Through outlier detection, data verification, and prediction model generation, the system has the characteristics of intelligence, self-adaptation, and efficient response, providing an intelligent solution for tunnel lighting.

[0004] Among the existing patents, the patent with publication number CN113840432A discloses "A method for adaptive timing control of tunnel intelligent lighting", including: 1. Establishing an adaptive timing control system for tunnel intelligent lighting; 2. Collecting environmental data outside the tunnel entrance; 3. Real-time control of tunnel intelligent lighting and accumulation of environmental data outside the tunnel entrance; 4. Constructing an adaptive timing control model for tunnel intelligent lighting and updating the model; 5. Adaptive timing control of tunnel intelligent lighting. The timing control method disclosed in this patent uses multi-dimensional information such as brightness, traffic flow, and vehicle speed outside the tunnel to establish or update a timing dimming model, and can still adaptively control the lighting in the event of damaged equipment in the environmental collection mechanism. This method can achieve the effect of improving lighting and saving lighting energy consumption, but the method has the following defects: 1. Limited model generalization ability The model of this method is only applicable to a single tunnel, and does not fully consider the migration and reuse of data in similar environments. For example, for adjacent tunnels with similar traffic and environmental conditions on a highway, their data and models cannot be shared, resulting in the need to independently establish a complete equipment system and dimming model for each tunnel, which not only increases the complexity of data storage and management, but also significantly increases hardware investment and operating costs, limiting the promotion and application of the system.

[0005] 2. Lack of data cleaning steps Failure to clean and process the collected data and directly using potentially noisy or erroneous data for model training can easily lead to reduced model performance and increase the risk of overfitting, thus affecting the actual operation of the system.

[0006] 3. Equipment abnormality without early warning function For abnormal collection equipment, the corresponding early warning module is not associated, and real-time detection and fault reporting cannot be achieved, which is not conducive to timely maintenance of equipment and long-term stable operation of the system. Summary of the invention

[0007] The main purpose of the present invention is to provide a tunnel timing dimming system that integrates sensor data and data migration, solve the problems existing in the prior art, ensure the authenticity and reliability of data, improve the generalization ability of the model, facilitate inspection and maintenance, and effectively utilize data to reduce sensor deployment costs.

[0008] In order to achieve the above object, the solution of the present invention is: A tunnel sequential dimming system integrating sensor data and data migration, comprising a sensor module, a data storage module, a data processing and verification module, a dimming module, a lighting control module, a data monitoring and alarm module and a data interface module; The sensor module is used to monitor data and collect the data required for dimming. ; The data storage module is used to perform data classification and data storage functions; data classification will dim the required data Marked as hole brightness , precipitation , Cloud Cover , Illuminance , Traffic flow , vehicle speed , Astronomical time and holiday labels Eight data categories; data storage includes a local storage unit and a cloud storage unit, and stores a tunnel basic information table; The data processing and verification module includes an abnormality detection unit and a data correction unit; the abnormality detection unit identifies the data required for dimming through an algorithm. The abnormal values ​​in the data are detected and an early warning is issued, and the data correction unit corrects or discards the abnormal values ​​through the dimming model; The dimming module includes a dimming strategy and a dimming model; the dimming strategy converts the data required for dimming into Combined with the tunnel basic information in the tunnel basic information table, the required brightness of different areas of the tunnel is calculated, and intelligent dimming is achieved through the lighting control module; the dimming model is based on the long short-term memory network, which is divided into the following according to the data source: the first model - the error correction LSTM model based on historical monitoring data, the second model - the real-time dimming model based on monitoring data, the third model - the LSTM time series dimming model based on historical monitoring data, and the fourth model - the LSTM time series dimming model based on migration data; When the collected data exceeds 1 month, the first model is trained: the first model includes sub-models corresponding to the number of data categories of the data storage module, and the input of the sub-model is , the input feature is the outlier , the dimension of the input data is , the output of the sub-model is an outlier Corrected forecast value , the output data dimension is ;in, Represents the data category, and its value is ; Represents the number of samples, whose value is the total amount of data and time step of the corresponding data category The difference between The value of is the product of the number of days for data collection and the monitoring frequency; When a certain type of data is judged as an outlier After that, the data correction unit calls the sub-model related to this type of data in the first model to correct the abnormal value. Input the corresponding sub-model and predict the predicted value , and then the predicted value Substitute the other sub-models in the first model to predict and get the prediction results. If the prediction results are consistent with the data required for dimming If the difference between the predicted value and the predicted value does not exceed the preset prediction threshold, the predicted value is determined to be In line with historical laws, use predicted values Replace the data required for dimming Otherwise, discard the predicted value ; The second model acquires data required for dimming based on the sensor module or the fourth model, and calculates the required brightness of different areas of the tunnel; The third model uses historical monitoring data to learn and calculate the required brightness of different areas of the tunnel when the sensor module fails or is turned off; When the two tunnels have the same information, data migration is performed, and the historical monitoring data of the tunnel after migration and the migration data constitute a data set to train the fourth model, and the fourth model predicts the data required for dimming. ; The data monitoring and alarm module is used to monitor the operating status of the lighting system and issue alarms and record logs when abnormalities occur; The data interface module is used to connect to other traffic management systems and allow administrators to remotely control them.

[0009] The sensor module includes a tunnel entrance brightness meter, a precipitation monitor, an all-sky imager, an illuminance meter, a vehicle flow and speed monitor, and an astronomical timer; the tunnel entrance brightness meter is used to monitor the light intensity at the tunnel entrance, the precipitation monitor is used to monitor whether it is raining and the amount of precipitation, the all-sky imager is used to monitor the cloudiness of the sky, the illuminance meter is used to monitor the illumination, the vehicle flow and speed monitor is used to monitor the vehicle flow and speed, and the astronomical timer is used to record astronomical time.

[0010] The data storage module adopts an SQL database; the tunnel basic information table stores the length, design speed and number of lanes of each tunnel, as well as the location of the tunnel on the map.

[0011] The abnormality detection unit uses the IQR method to detect abnormal points or erroneous data. The detection value of a certain type of data is higher than the third quartile or below the first quartile It is judged as an outlier when , and its calculation formula is: ; ; in, , They represent the lower limit and upper limit of a certain type of data when it is judged as a normal value. Represents the multiplication factor.

[0012] In the data processing and verification module, when an abnormal value is detected After that, the anomaly detection unit will detect the abnormal value Mark and save to the local storage unit so that the abnormal value can be called by the data correction unit .

[0013] The dimming formula of the dimming strategy in the dimming module is as follows: ; in, Represents the brightness outside the cave; , represents the brightness of entrance segments 1 and 2 respectively; Representatives and traffic volume , vehicle speed The relevant function, whose value determines the brightness reduction factor at the entrance section, is taken according to Table 4.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); , , Represent the brightness of transition sections 1, 2, and 3 respectively; Represents the brightness of the middle segment; Representatives and traffic volume , vehicle speed The relevant functions are taken according to Table 6.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); Represents the brightness of the exit section; Represents and tunnel length The relevant functions shall be taken according to Tables 7.0.1 and 7.0.2 of the "Highway Tunnel Lighting Specifications" (JTG / T D70 / 2-01—2014).

[0014] The dimming module is preset with a dimming threshold. The system executes the dimming strategy only when the difference between the brightness value calculated at a certain moment and the previous brightness value exceeds the dimming threshold. In the fourth model, judgment is made based on the weather, entrance brightness and traffic flow information of the tunnel, and a judgment threshold is set. When the difference of each information does not exceed the preset judgment threshold, it is considered that the two tunnels have the same information. The dimming module is preset with a data volume threshold or a time threshold. When the newly added data volume exceeds the data volume threshold or the accumulated time exceeds the time threshold, the dimming model is retrained and saved in the local storage unit of each tunnel.

[0015] The lighting control module includes a dimming controller, a regional lighting control unit, a brightness gradient control module and an emergency lighting control unit; the dimming controller executes a dimming strategy; the regional lighting control unit adjusts the tunnel lighting intensity in different regions; the brightness gradient control module realizes a gradual transition of lighting; and the emergency lighting control unit starts emergency lighting in the event of system abnormality or emergency.

[0016] The data monitoring and alarm module includes a status monitoring unit, an alarm system and a log storage and backtracking module; the status monitoring unit is used to monitor the system operation status in real time; the alarm system issues an alarm and records a log when the system detects an abnormality or failure; the log storage and backtracking module is used to record all system operations and abnormal situations.

[0017] The data interface module includes an external system interface and a remote control interface; the external system interface is connected to other traffic management systems to achieve data sharing and linkage control; the remote control interface allows administrators to remotely adjust system parameters and monitor system status.

[0018] After adopting the above technical solution, the present invention constructs a more perfect tunnel timing dimming system: 1. Added data processing and verification modules to judge, correct, and fill in the detection data to ensure that the data is authentic and reliable, thereby improving the generalization ability of the dimming model; 2. Add data monitoring and alarm modules to judge the status of sensors, issue equipment abnormality reports, and remind maintenance personnel to perform maintenance or replacement; 3. A migration strategy has been added so that tunnels with similar environment, traffic flow and other information can rely on the established dimming model for migration training to generate a timing dimming model suitable for the tunnel, thereby effectively utilizing data and reducing sensor deployment to reduce costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a module schematic diagram of a specific embodiment of the present invention.

[0020] Figure 2 FIG. 4 is a schematic diagram of a dimming strategy according to a specific embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to further explain the technical solution of the present invention, the present invention is described in detail below through specific embodiments.

[0022] refer to Figure 1 As shown, the present invention discloses a tunnel sequential dimming system integrating sensor data and data migration, including a sensor module, a data storage module, a data processing and verification module, a dimming module, a lighting control module, a data monitoring and alarm module and a data interface module; The sensor module is used to monitor data and collect the data required for dimming. ; The data storage module is used to perform the functions of data classification and data storage; data classification will dim the required data Marked as hole brightness , precipitation , Cloud Cover , Illuminance , Traffic flow , vehicle speed , Astronomical time and holiday labels Eight data categories; data storage includes a local storage unit and a cloud storage unit, and stores a tunnel basic information table; The data processing and verification module includes an abnormality detection unit and a data correction unit; the abnormality detection unit uses an algorithm to identify the data required for dimming The abnormal values ​​in the data are detected and an early warning is issued, and the data correction unit corrects or discards the abnormal values ​​through the dimming model; The dimming module includes a dimming strategy and a dimming model; the dimming strategy converts the data required for dimming into Combined with the tunnel basic information in the tunnel basic information table, the required brightness of different areas of the tunnel is calculated, and intelligent dimming is achieved through the lighting control module; the basis of the dimming model is the long short-term memory network (LSTM, Long Short-Term Memory, is a special recursive neural network that is particularly suitable for processing and predicting time series data. LSTM introduces memory units and gating mechanisms to solve the gradient vanishing and explosion problems of traditional recursive neural networks when processing long sequences). According to the data source, it is divided into: the first model - the error correction LSTM model based on historical monitoring data, the second model - the real-time dimming model based on monitoring data, the third model - the LSTM time series dimming model based on historical monitoring data, and the fourth model - the LSTM time series dimming model based on migration data; When the collected data exceeds 1 month, the first model is trained: the first model includes sub-models corresponding to the number of data categories of the data storage module, and the input of the sub-model is , the input feature is the outlier , the dimension of the input data is , the output of the sub-model is an outlier Corrected forecast value , the output data dimension is ;in, Represents the data category, and its value is ; Represents the number of samples, whose value is the total amount of data and time step of the corresponding data category The difference between The value is the product of the number of days for data collection and the monitoring frequency (times / day); When a certain type of data is judged as an outlier After that, the data correction unit calls the sub-model related to this type of data in the first model to correct the abnormal value. Input the corresponding sub-model and predict the predicted value , and then the predicted value Substitute the other sub-models in the first model to predict and get the prediction results. If the prediction results are consistent with the data required for dimming If the difference between the actual monitoring effect of the sensor module and the predicted value does not exceed the preset prediction threshold, the predicted value is determined to be In line with historical laws, use predicted values Replace the data required for dimming Otherwise, discard the predicted value ; The second model obtains the data required for dimming based on the sensor module or the fourth model, and calculates the required brightness of different areas of the tunnel; The third model uses historical monitoring data for learning when the sensor module fails or is turned off (see Example 1 below for details) and calculates the required brightness of different areas of the tunnel; When two tunnels have the same information, data migration is performed (see Example 2 below for details), and the historical monitoring data of the tunnel after migration and the migration data constitute a data set to train the fourth model, and the fourth model predicts the data required for dimming. ; The data monitoring and alarm module is used to monitor the operating status of the lighting system and issue alarms and record logs when abnormalities occur; The data interface module is used to connect to other traffic management systems and allow administrators to remotely control them.

[0023] Through the above scheme, the present invention constructs a more perfect tunnel timing dimming system in order to overcome the above existing technical defects: 1. Added data processing and verification modules to judge, correct, and fill in the detection data to ensure that the data is authentic and reliable, thereby improving the generalization ability of the dimming model; 2. Add data monitoring and alarm modules to judge the status of sensors, issue equipment abnormality reports, and remind maintenance personnel to perform maintenance or replacement; 3. A migration strategy has been added so that tunnels with similar environment, traffic flow and other information can rely on the established dimming model for migration training to generate a timing dimming model suitable for the tunnel, thereby effectively utilizing data and reducing sensor deployment to reduce costs.

[0024] The specific implementation details of the above tunnel sequential dimming system are shown below.

[0025] The above-mentioned sensor modules include a tunnel entrance brightness meter, a precipitation monitor, a full sky imager, a illuminance meter, a vehicle flow and speed monitor and an astronomical timer; the tunnel entrance brightness meter is used to monitor the light intensity of the tunnel entrance, the precipitation monitor is used to monitor whether it is raining and the amount of precipitation, the full sky imager is used to monitor the cloudiness of the sky (i.e., monitor the sky state, whether it is sunny, cloudy, etc.), the illuminance meter is used to monitor the illumination, the vehicle flow and speed monitor is used to monitor the vehicle flow and speed, and the astronomical timer is used to record the astronomical time.

[0026] The local storage unit is used to temporarily store processed data. It can store a certain amount of data to avoid data loss due to network disconnection. After the network is restored, the data can be uploaded to the cloud. The models used by the lighting system can also be saved in the local storage unit. The data storage module uses an SQL database. In addition to saving basic information such as the length, design speed, and number of lanes of each tunnel, the tunnel basic information table should also include the location of each tunnel on the map, so that the data collected from other similar tunnels can be migrated to the target tunnel later.

[0027] The above abnormality detection unit uses the IQR (interquartile range) method to detect abnormal points or erroneous data. The detection value of a certain type of data is higher than the third quartile or below the first quartile It is judged as an outlier when , and its calculation formula is: ; ; in, , They represent the lower limit and upper limit of a certain type of data when it is judged as a normal value. Represents the multiplication factor ( Can be set to 1.5. When the detection value is and If the value is within the range, it is considered normal; otherwise, the data is considered abnormal); In the above data processing and verification module, when an abnormal value is detected After that, the anomaly detection unit will detect the abnormal value Mark and save to the local storage unit so that the abnormal value can be called by the data correction unit .

[0028] The dimming formula of the dimming strategy in the above dimming module is as follows: ; in, Represents the brightness outside the cave; , represents the brightness of entrance segments 1 and 2 respectively; Representatives and traffic volume (Unit: veh / (h.ln)), vehicle speed (Unit: km / h) related function, whose value determines the brightness reduction factor at the entrance section, and is determined according to Table 4.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); , , Represent the brightness of transition sections 1, 2, and 3 respectively; Represents the brightness of the middle segment; Representatives and traffic volume , vehicle speed The relevant functions are taken according to Table 6.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); Represents the brightness of the exit section; Represents and tunnel length The relevant functions shall be taken according to Tables 7.0.1 and 7.0.2 of the "Highway Tunnel Lighting Specifications" (JTG / T D70 / 2-01—2014).

[0029] Furthermore, a dimming threshold is preset in the dimming module. The system executes the dimming strategy only when the difference between the brightness value calculated at a certain moment and the previous brightness value exceeds the dimming threshold, which can avoid the reduction of the service life of the LED lamp due to excessive dimming during use.

[0030] At the same time, when the tunnel is one-way traffic, the interpolation edge points are as follows: Time Press calculate; Time Press calculate; Time Press calculate; Time Press Calculate; then , , , , , , , , , ; When the tunnel is a two-way traffic tunnel, the interpolation edge points are as follows: Time Press calculate; Time Press calculate; Time Press calculate; Time Press Calculate; then , , , , , , , , , .

[0031] In the fourth model mentioned above, judgment is made based on information such as tunnel weather, entrance brightness, and vehicle flow, and a judgment threshold is set. When the difference in each piece of information does not exceed the preset judgment threshold, the two tunnels are considered to have the same information. For example, if Tunnel 1 and Tunnel 2 have similar weather, entrance brightness, and vehicle flow, and Tunnel 1 is equipped with a sensor module, while Tunnel 2 is not equipped with a sensor module or only a part of it is equipped, the data information of Tunnel 1 can be migrated to Tunnel 2 through certain corrections. This migration strategy allows the lighting system to operate normally even when the sensor module of Tunnel 2 lacks a certain monitoring item (such as the entrance brightness meter is not arranged or fails); and, during data migration, if Tunnel 2 has arranged a certain monitoring item, such as a precipitation monitor, the monitoring data of Tunnel 2 is used. Migration data with tunnel 1 (except Other items except Figure 2 .

[0032] The dimming module is preset with a data volume threshold or a time threshold. When the newly added data volume exceeds the data volume threshold or the accumulated time exceeds the time threshold, the dimming model is retrained and saved in the local storage unit of each tunnel.

[0033] The above-mentioned lighting control module includes a dimming controller, a regional lighting control unit, a brightness gradient control module and an emergency lighting control unit; the dimming controller executes a dimming strategy to control the brightness change of the light; the regional lighting control unit adjusts the tunnel lighting intensity in different areas to ensure appropriate brightness in different areas; the brightness gradient control module realizes a gradual transition of lighting to prevent sudden changes in brightness from causing driving discomfort; the emergency lighting control unit starts the emergency lighting in the event of system abnormality or emergency.

[0034] The above-mentioned data monitoring and alarm module includes a status monitoring unit, an alarm system and a log storage and backtracking module; the status monitoring unit is used to monitor the system operation status in real time to ensure the normal operation of sensors and controllers; the alarm system issues an alarm and records a log when the system detects an abnormality or failure; the log storage and backtracking module is used to record all system operations and abnormal conditions to facilitate subsequent maintenance and fault analysis.

[0035] The above-mentioned data interface module includes an external system interface and a remote control interface; the external system interface is connected to other traffic management systems to realize data sharing and linkage control; the remote control interface allows the administrator to remotely adjust system parameters and monitor system status.

[0036] Two specific embodiments of the present invention are shown below.

[0037] The background is as follows: (1) On a certain expressway, there are two adjacent tunnels, divided into tunnels A and B according to the direction of traffic, 3 km apart, with no entrance, exit or rest area in between. A complete sensor module is deployed in tunnel A, and a partial sensor module is deployed in tunnel B.

[0038] (2) The complete sensor module includes: cave entrance brightness meter, precipitation monitor, all-sky imager, illuminance meter, traffic flow and speed monitor and astronomical timer.

[0039] (3) The data collected from tunnel A are marked as the brightness of the tunnel entrance. , precipitation , Cloud Cover , Illuminance , Traffic flow , vehicle speed , Astronomical time and holiday labels , get the data needed for dimming .

[0040] Embodiment 1: 1-1. Establish a dimming system in Tunnel A: The system includes a monitoring host, a tunnel entrance brightness meter, a precipitation monitor, an all-sky imager, a illuminance meter, a vehicle flow and speed monitor, an astronomical timer, and a lighting unit associated with the dimming system in the tunnel; the lighting unit should include: a microcontroller module (as a core processing unit), a bus module (connecting the microcontroller module and the monitoring host), a dimming lighting lamp (the microcontroller module outputs a control signal for brightness adjustment), and the microcontroller module communicates with the monitoring host through the bus module.

[0041] 1-2. Tunnel division: According to the "Highway Tunnel Lighting Design Rules" (JTG / T DXX-2010), Tunnel A is divided into the entrance section (TH1), the entrance stage (TH2), the first transition section (TR1), the second transition section (TR2), the third transition section (TR3), the middle section (IN), the exit connection section (EX1), and the exit section (EX2).

[0042] The length of each segment is calculated according to the following formula: ; ; ; ; ; in, Represents the illuminated parking sight distance, in meters; Represents the clear height in the tunnel, in m; Represents the design speed of the tunnel.

[0043] 1-3. Data collection: Pack the data collected at the same time into the data set of tunnel A , , at this time the data processing and verification module will to verify.

[0044] 1-4. When the brightness of the hole or When the data is abnormal, it will be marked as abnormal and recorded in the data monitoring and alarm module; then the system will call the third model to predict the data and get the predicted value ; Then the predicted value As the true value, substitute other factors predicted by the third model (such as predicted precipitation ) is used to make predictions in the sub-model, and the predicted value ( ) and the actual monitoring value ( ) is within the threshold, then recurse to the next ( ) prediction sub-model, and so on, if the prediction results are all within the threshold, then the data ( ) of the predicted value ( ) conforms to the rules, complete it, otherwise discard the data.

[0045] 1-5. When the monitoring data of a certain sensor (hole luminance meter) is abnormal continuously and exceeds the threshold (5 times), or multiple sensors are abnormal at the same time (3 or more), the data monitoring and alarm module will respond, report the abnormal information to relevant personnel, and automatically switch to the traditional dimming mode. Subsequent staff can manually switch to restore to the timing dimming model.

[0046] 1-6. Use corrected data ( ) replace the original abnormal data to form a corrected data set , calculate the brightness value of each segment according to the dimming formula , then the lighting control module controls each lighting unit to dim; in addition, when the difference between the newly calculated brightness value and the brightness value of the previous stage exceeds the dimming threshold, the system issues a dimming command through the lighting control module, thereby avoiding excessive dimming and reducing the service life of the lighting unit. With brightness value It will be stored in the data storage module as the basic data for subsequent data migration and LSTM model training.

[0047] 1-7. When there is more than 1 month of data, the first model can be trained. When there is more than 1 year of stored data, the third model can be trained.

[0048] 1-8. The operator can choose to turn off the sensor module and use the third model for time-series dimming. For example, if the dimming value for the next 30 days needs to be predicted, the number of features input to the LSTM model is 9 ( ), the input data is: time step is 7*monitoring frequency (times / day); number of samples: total data volume-time step; feature data, data dimension is (number of samples, time step, 9). The output data dimension is: (number of samples, 30*monitoring frequency, 1), that is, the dimming forecast value for the next 30 days is output. The system dims according to the corresponding time and the forecast value. If the difference between the current brightness and the calculated dimming value exceeds the dimming threshold, dimming is performed. In this case, the dimming threshold can be appropriately increased to reduce the sensitivity of the system.

[0049] Embodiment 2: 2-1. For tunnel B, its data is Indicates that the subscripts have the same meaning; Tunnel B is 3km away from Tunnel A, there is no entrance or exit between AB and they are on the same line, the two have the same designed speed and similar weather conditions.

[0050] 2-2. No opening brightness meter is arranged, and the opening brightness adopts the design opening brightness ratio The product of the monitoring value of tunnel A ; No precipitation monitoring equipment is deployed, ; No all-sky imaging instrument is deployed, ; No traffic flow and speed monitoring equipment is installed, and the traffic flow With speed , the corresponding data should be shifted in time; no astronomical timer is arranged, ; Same label for holidays ; Arrange illuminance meter, measure .

[0051] 2-3. Through 2-2, we obtained the data set of tunnel B .

[0052] 2-4. Referring to 1-4, 1-5, 1-6 of Example 1, contrast Perform abnormality determination, correction and data storage.

[0053] 2-5. Referring to 1-7 of Example 1, model training of Tunnel B is performed.

[0054] 2-6. Referring to 1-8 of Example 1, the time series dimming system can be manually turned on to perform time series prediction on tunnel B. The model at this time is the fourth model, that is, the LSTM prediction model based on migration data.

[0055] It should be noted that if the two tunnels do not have so many similarities, sensors can be deployed separately for the factors that cannot be measured using migration, and other data can be supplemented with migration strategies. The fourth model does not have strict restrictions on the number of features of the migration data, and can be flexibly adjusted according to specific circumstances. This helps to ensure data accuracy and reduce costs to a certain extent.

[0056] The above embodiments and drawings do not limit the product form and style of the present invention. Any appropriate changes or modifications made by ordinary technicians in the relevant technical field should be deemed to be within the patent scope of the present invention.

Claims

1. A tunnel sequential dimming system integrating sensor data and data migration, characterized in that: It includes sensor module, data storage module, data processing and verification module, dimming module, lighting control module, data monitoring and alarm module and data interface module; The sensor module is used to monitor data and collect the data required for dimming. ; The data storage module is used to perform data classification and data storage functions; data classification will dim the required data Marked as hole brightness , precipitation , Cloud Cover , Illuminance , Traffic flow , vehicle speed , Astronomical time and holiday labels Eight data categories; data storage includes a local storage unit and a cloud storage unit, and stores a tunnel basic information table; The data processing and verification module includes an abnormality detection unit and a data correction unit; the abnormality detection unit identifies the data required for dimming through an algorithm. The abnormal values ​​in the data are detected and an early warning is issued, and the data correction unit corrects or discards the abnormal values ​​through the dimming model; The dimming module includes a dimming strategy and a dimming model; the dimming strategy converts the data required for dimming into Combined with the tunnel basic information in the tunnel basic information table, the required brightness of different areas of the tunnel is calculated, and intelligent dimming is achieved through the lighting control module; the dimming model is based on the long short-term memory network, which is divided into the following according to the data source: the first model - the error correction LSTM model based on historical monitoring data, the second model - the real-time dimming model based on monitoring data, the third model - the LSTM time series dimming model based on historical monitoring data, and the fourth model - the LSTM time series dimming model based on migration data; When the collected data exceeds 1 month, the first model is trained: the first model includes sub-models corresponding to the number of data categories of the data storage module, and the input of the sub-model is , the input feature is the outlier , the dimension of the input data is , the output of the sub-model is an outlier Corrected forecast value , the output data dimension is ;in, Represents the data category, and its value is ; Represents the number of samples, whose value is the total amount of data and time step of the corresponding data category The difference between The value of is the product of the number of days for data collection and the monitoring frequency; When a certain type of data is judged as an outlier After that, the data correction unit calls the sub-model related to this type of data in the first model to correct the abnormal value. Input the corresponding sub-model and predict the predicted value , and then the predicted value Substitute the other sub-models in the first model to predict and get the prediction results. If the prediction results are consistent with the data required for dimming If the difference between the predicted value and the predicted value does not exceed the preset prediction threshold, the predicted value is determined to be In line with historical laws, use predicted values Replace the data required for dimming Otherwise, discard the predicted value ; The second model acquires data required for dimming based on the sensor module or the fourth model, and calculates the required brightness of different areas of the tunnel; The third model uses historical monitoring data to learn and calculate the required brightness of different areas of the tunnel when the sensor module fails or is turned off; When two tunnels have the same information, data migration is performed, and the historical monitoring data of the tunnel after migration and the migration data constitute a data set to train the fourth model, and the fourth model predicts the data required for dimming. ; The data monitoring and alarm module is used to monitor the operating status of the lighting system and issue alarms and record logs when abnormalities occur; The data interface module is used to connect to other traffic management systems and allow administrators to remotely control them.

2. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The sensor module includes a tunnel entrance brightness meter, a precipitation monitor, an all-sky imager, an illuminance meter, a vehicle flow and speed monitor, and an astronomical timer; the tunnel entrance brightness meter is used to monitor the light intensity at the tunnel entrance, the precipitation monitor is used to monitor whether it is raining and the amount of precipitation, the all-sky imager is used to monitor the cloudiness of the sky, the illuminance meter is used to monitor the illumination, the vehicle flow and speed monitor is used to monitor the vehicle flow and speed, and the astronomical timer is used to record astronomical time.

3. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The data storage module adopts SQL database; The tunnel basic information table stores the length, design speed, number of lanes of each tunnel, and the location of the tunnel on the map.

4. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The abnormality detection unit uses the IQR method to detect abnormal points or erroneous data. The detection value of a certain type of data is higher than the third quartile or below the first quartile It is judged as an outlier when , and its calculation formula is: ； ； in, , They represent the lower limit and upper limit of a certain type of data when it is judged as a normal value. Represents the multiplication factor.

5. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: In the data processing and verification module, when an abnormal value is detected After that, the anomaly detection unit will detect the abnormal value Mark and save to the local storage unit so that the abnormal value can be called by the data correction unit .

6. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The dimming formula of the dimming strategy in the dimming module is as follows: ； in, Represents the brightness outside the cave; , represents the brightness of entrance segments 1 and 2 respectively; Representatives and traffic volume , vehicle speed The relevant function, whose value determines the brightness reduction factor at the entrance section, is taken according to Table 4.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); , , Represent the brightness of transition sections 1, 2, and 3 respectively; Represents the brightness of the middle segment; Representatives and traffic volume , vehicle speed The relevant functions are taken according to Table 6.1.1 of the "Highway Tunnel Lighting Regulations" (JTG / T D70 / 2-01—2014); Represents the brightness of the exit section; Represents and tunnel length The relevant functions shall be taken according to Tables 7.0.1 and 7.0.2 of the "Highway Tunnel Lighting Specifications" (JTG / T D70 / 2-01—2014).

7. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The dimming module is preset with a dimming threshold. The system executes the dimming strategy only when the difference between the brightness value calculated at a certain moment and the previous brightness value exceeds the dimming threshold. In the fourth model, judgment is made based on the weather, entrance brightness and traffic flow information of the tunnel, and a judgment threshold is set. When the difference of each information does not exceed the preset judgment threshold, it is considered that the two tunnels have the same information. The dimming module is preset with a data volume threshold or a time threshold. When the newly added data volume exceeds the data volume threshold or the accumulated time exceeds the time threshold, the dimming model is retrained and saved in the local storage unit of each tunnel.

8. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The lighting control module includes a dimming controller, a regional lighting control unit, a brightness gradient control module and an emergency lighting control unit; the dimming controller executes a dimming strategy; the regional lighting control unit adjusts the tunnel lighting intensity in different regions; the brightness gradient control module realizes a gradual transition of lighting; The emergency lighting control unit activates the emergency lighting in case of system abnormality or emergency.

9. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The data monitoring and alarm module includes a status monitoring unit, an alarm system and a log storage and backtracking module; the status monitoring unit is used to monitor the system operation status in real time; the alarm system issues an alarm and records a log when the system detects an abnormality or failure; the log storage and backtracking module is used to record all system operations and abnormal situations.

10. The tunnel sequential dimming system integrating sensor data and data migration as claimed in claim 1, characterized in that: The data interface module includes an external system interface and a remote control interface; the external system interface is connected to other traffic management systems to achieve data sharing and linkage control; the remote control interface allows administrators to remotely adjust system parameters and monitor system status.

Images