Bullet train automatic light control system based on computer vision assistance

Abstract

The invention discloses a motor car automatic light control system based on computer vision assistance, and relates to the technical field of motor car automatic light control, and the system comprises a severe weather identification module, a light dynamic adjustment module and an intelligent control and feedback module. By arranging the multi-position high-definition camera outside the bullet train and applying a severe weather feature extraction algorithm, precise recognition of severe weather such as rainstorm, violent snow and dense fog is achieved, the problem that the severe weather recognition capacity of an existing system is limited is solved, and by means of a self-adaptive light regulation and control algorithm and a differential adjustment strategy for different types of lamplight, the recognition accuracy of the bullet train is improved. Light parameters are dynamically adjusted according to severe weather and ambient light intensity, contrast is enhanced, color temperature is adjusted, a comfortable light environment is created, visual experience of passengers is improved, sight of a driver is prevented from being interfered, information such as weather and running speed and geographic position of the bullet train is fused through an environment information fusion module, comprehensive and reasonable light adjustment is achieved, and the service life of the bullet train is prolonged. And the intelligent operation requirement of the bullet train is met.

Description

Technical Field

[0001] The invention relates to the technical field of automatic light control for motor vehicles, and in particular to an automatic light control system for motor vehicles based on computer vision assistance. Background Art

[0002] EMU, or multiple-unit train, is a train set consisting of several powered and non-powered vehicles. Relying on advanced power distribution technology, it has a higher operating speed, better acceleration performance and stronger climbing ability compared to traditional trains, which greatly improves the efficiency of railway transportation and the travel experience of passengers. It has become an important part of modern railway transportation. In the railway transportation management system, the EMU automatic lighting control system is a key technology for automatically adjusting the interior lighting of the EMU. Its core purpose is to automatically and accurately adjust the brightness, color temperature, contrast and other parameters of the interior lighting according to various actual conditions during the operation of the EMU, such as different operating time periods, changes in external light and special weather conditions, so as to create a comfortable and suitable riding light environment for passengers, while ensuring that the driver has a clear field of vision in different environments, which helps to improve driving safety.

[0003] However, there are many problems with the existing automatic light control system for EMUs. When facing bad weather, the recognition ability is extremely limited. Traditional systems often rely only on simple light sensors to perceive changes in external light, and cannot accurately distinguish different bad weather conditions such as heavy rain, heavy snow, and dense fog, which makes it difficult to make targeted light adjustments in bad weather. For example, in heavy rain, the scattering and reflection of light by raindrops will cause the external light to change dramatically, and the existing system cannot effectively enhance the light contrast inside the car, making it difficult for passengers to see the signs and objects inside the car, and the driver is also easily affected by light interference and affected vision. In dense fog, the existing system cannot accurately adjust the light color temperature to a level close to natural light. Passengers are prone to visual fatigue when they are in uncomfortable light for a long time, and it is difficult for the driver to clearly observe the external road conditions. In addition, the existing system has obvious deficiencies in the coordination with other EMU systems, and cannot be effectively integrated with information such as the EMU running speed and geographical location. It cannot make comprehensive and reasonable light adjustments according to different operating scenarios, and cannot meet the growing demand for intelligent and humanized operation of EMUs. Therefore, it is necessary to propose an EMU automatic light control system based on computer vision to solve the problems in the existing technology. Summary of the invention

[0004] The purpose of the present invention is to make up for the shortcomings of the prior art and provide an automatic light control system for trains based on computer vision assistance. It can improve the comfort of passengers riding in trains in bad weather by real-time monitoring of bad weather and intelligently and dynamically adjusting the light inside the train, while ensuring that the driver's vision is not disturbed and driving safety is ensured.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a computer vision-assisted EMU automatic light control system, the system comprising: a severe weather recognition module, a light dynamic adjustment module and an intelligent control and feedback module;

[0006] The severe weather recognition module is set outside the train and consists of high-definition cameras located at the front, both sides of the train body and the rear of the train. It is used to collect images of the train's surrounding environment in real time and uses a severe weather feature extraction algorithm. The calculation formula is: ,in Represents the severe weather characteristic value, is the corresponding image area The weight coefficient of is the feature enhancement factor, with a value range of 1.2-1.8. The algorithm is used to extract the key features of the image, and the classifier is used to determine the type of severe weather. The classifier is a weather classification model based on a convolutional neural network;

[0007] The light dynamic adjustment module is connected to the severe weather recognition module, and an adaptive light control algorithm is applied based on the recognition result. The algorithm is based on the formula ,in is the adjusted light parameter, is to adjust the front light parameters, is the ambient light intensity, , is the light adjustment factor, , To correct the parameters, the light parameters of the brightness, color temperature and contrast of the interior lights are adjusted accordingly;

[0008] The intelligent control and feedback module is connected to the severe weather recognition module and the light dynamic adjustment module, receives the severe weather recognition result and the data of the in-vehicle light sensor and the passenger feedback terminal, and uses the intelligent decision fusion algorithm to obtain the formula: ,in For decision-making instructions, is the decision factor weight, Based on the severe weather identification results , light sensor data , Passenger feedback data The decision function is used to optimize and control the light adjustment strategy.

[0009] Furthermore, the layout of the high-definition cameras in the severe weather recognition module is as follows: three are installed at the front of the vehicle in a fan-shaped distribution, with a viewing angle covering a 180° range directly in front; two are installed on each side of the vehicle body, each inclined at 30°, with a viewing angle covering an area where the angle between the side of the vehicle body and the direction of travel of the vehicle is 60°-150°; and one is installed at the rear of the vehicle, with a viewing angle covering a 120° range directly behind.

[0010] Furthermore, the weight coefficient in the severe weather feature extraction algorithm is Through the analysis of historical image data, it is determined that for heavy rain weather, the image area weight of raindrops accounts for 60% of the total weight, of which the weights of raindrop size, speed, and distribution area are 20%, 20%, and 20% respectively. In heavy snow weather, the image area weights of snowflake shape, falling speed, and coverage degree account for 20%, 30%, and 30% of the total weight respectively. In dense fog weather, the image area weights of fog concentration and distribution range account for 40% and 40% of the total weight respectively, and the remaining 20% ​​of the weight is allocated to other feature areas.

[0011] Furthermore, the light adjustment coefficient in the light dynamic adjustment module and The fuzzy logic reasoning system determines the current bad weather type, ambient light intensity and preset light standards in the car. The bad weather types are divided into three categories: heavy rain, heavy snow and dense fog. The ambient light intensity is divided into three intervals: low, medium and high, corresponding to 0-100lux, 101-500lux, 501lux and above. The preset light standards in the car are set to brightness 400-600lux, color temperature 4000K-6000K, and contrast 50%-70%. The fuzzy logic reasoning system calculates these variables through specific fuzzy rules. and If the weather type is heavy rain and the ambient light intensity is high, then set .7, .3.

[0012] Furthermore, the adjustment strategies of the dynamic light adjustment module for different types of interior lights are as follows: the brightness adjustment range of the reading light is within ±10%, the color temperature adjustment range is 4500K-5500K, the brightness of the aisle light is increased by 20% in bad weather, and the contrast is increased by 15%. The ceiling light of the car adjusts its own light parameters according to the adjustment of the reading light and aisle light, with the brightness being 1.2 times the average brightness of the reading light and the aisle light, the color temperature being the average color temperature of the two, and the contrast being the average contrast of the two.

[0013] Furthermore, the decision factor weights in the intelligent control and feedback module , the neural network algorithm uses the running time and mileage of the train as input variables for self-learning and updating. A three-layer fully connected neural network is used, with the input layer being the running time and mileage, and the output layer being the decision factor weights. , with each running 1000 kilometers and 100 hours as an update cycle, calculate the mean square error between the output data of each module and the preset ideal data under different decision instructions, and update it according to the gradient descent method , so that the system can adapt to the actual operation changes of the train and optimize the light adjustment strategy.

[0014] Furthermore, the intelligent control and feedback module has fault diagnosis and emergency processing functions. When abnormalities are detected in the severe weather recognition module, the light dynamic adjustment module and the internal data transmission of the system, the intelligent control and feedback module starts the fault diagnosis program. The program quickly locates the fault point by analyzing the rationality of the output data of each module and comparing it with the preset normal operating parameter range, and starts corresponding emergency processing measures for different fault types.

[0015] Furthermore, the system also includes an environmental information fusion module, which integrates the weather information obtained by the severe weather identification module with the running speed and geographical location information of the train. The running speed is divided into three intervals: low speed, medium speed and high speed. The geographical location information is obtained by the on-board GPS, including longitude and latitude, altitude and terrain type. The terrain type classification standard is: altitude > 500 meters is mountainous area, slope > 5% is hilly area. According to the combination of weather type, running speed interval and terrain type, control instructions for the dynamic light adjustment module are generated according to the following rules: when it is raining heavily in mountainous areas and driving at low speed, the light brightness is increased by 30% and the contrast is increased by 20%; when it is snowing heavily in plains and driving at high speed, the color temperature is reduced by 500K and the brightness is reduced by 10%, so that the light adjustment fits the actual operation scene.

[0016] Furthermore, the energy-saving optimization strategy of the dynamic light adjustment module is: when the ambient light intensity change rate is less than 0.2lux / s due to severe weather changes, the light adjustment range does not exceed ±5%, and the adjustment frequency is not higher than once every 5 minutes. The distribution of people is detected by infrared sensors installed on the top of the car and on both sides of the aisle, and the activity of people is judged by the principle of thermal imaging. When there is no activity in a certain area for 3 consecutive minutes, the brightness of the light in the area is reduced by 30%. If there is no activity for 10 consecutive minutes, the light in the area is turned off, thereby meeting the lighting needs in the car while reducing the energy consumption of the system.

[0017] Furthermore, the severe weather recognition module has an early warning function for extremely severe weather. The early warning function is implemented as follows: the upper limit of the normal severe weather range is the 95th percentile of the W value in the historical data, and the severe weather characteristic value is When the intensity exceeds 1.5 times the upper limit of the normal severe weather range, it is determined that extremely severe weather may occur. The severe weather recognition module sends a warning signal to the EMU central control system and sends a special instruction to the light dynamic adjustment module. The light dynamic adjustment module increases the brightness by 30% and the contrast by 25% to cope with the visual environment in extremely severe weather.

[0018] Compared with the existing technology, the automatic light control system for electric vehicles based on computer vision assistance has the following beneficial effects:

[0019] 1. The present invention sets high-definition cameras at multiple positions outside the train and uses a severe weather feature extraction algorithm to achieve accurate recognition of severe weather such as heavy rain, heavy snow, and dense fog, thereby solving the problem of limited severe weather recognition capability of existing systems. With the help of an adaptive light control algorithm and differentiated adjustment strategies for different types of lights, the light parameters are dynamically adjusted according to severe weather and ambient light intensity, the contrast is enhanced, the color temperature is adjusted, a comfortable light environment is created, the visual experience of passengers is improved, and interference with the driver's vision is avoided. Through the environmental information fusion module, the weather is integrated with information such as the train's running speed and geographical location to achieve comprehensive and reasonable light adjustment to meet the needs of intelligent operation of trains.

[0020] 2. The present invention dynamically optimizes the light adjustment strategy through the intelligent decision-making fusion algorithm of the intelligent control and feedback module, combined with the results of severe weather identification, light sensor data and passenger feedback. It realizes refined management of light adjustment, can adjust the light in real time according to actual conditions, reduce unnecessary light adjustment operations, and reduce system energy consumption. At the same time, the self-learning update mechanism of the module can continuously adapt to various changes in the operation of the EMU, ensure the long-term stability and efficient operation of the system, extend the service life of the system, reduce maintenance costs, and further improve the overall performance and economy of the EMU automatic light control system.

[0021] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 It is a structural diagram of an automatic light control system for electric vehicles based on computer vision assistance;

[0024] Figure 2 This is a workflow diagram of a train automatic lighting control system based on computer vision assistance. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] Embodiment 1

[0027] On a winter day, a high-speed train full of passengers was traveling from one northern city to another. The route of the journey was complicated, passing through different terrain areas such as mountains and plains. The weather forecast showed that there would be snowfall and heavy fog in some areas of the route.

[0028] When the train just entered the mountainous section, the temperature outside the window gradually dropped and snowflakes began to fall. At this time, multiple high-definition cameras installed on the front, both sides and rear of the train quickly entered working state. The three cameras distributed in a fan shape on the front are like sharp eyes, with a field of view covering 180° directly in front, which can clearly capture large-area environmental images in the direction of the train's advance and perceive the road conditions and weather changes ahead in advance; two cameras tilted 30° on each side of the car body cleverly cover the area of ​​60°-150° between the side and the direction of advance, and are responsible for collecting detailed information on the side of the car body, including the angle and density of snowflakes falling, etc. The camera at the rear focuses on covering the 120° area directly behind to ensure that the situation behind the train can also be monitored in real time. These cameras collect images of the surrounding environment in all directions, providing a rich and comprehensive data foundation for subsequent severe weather identification.

[0029] The collected images are quickly transmitted to the severe weather recognition module, which uses a unique severe weather feature extraction algorithm To analyze images, for blizzard weather, after analyzing massive historical image data, the weights of the image areas related to snowflake shape, falling speed, and coverage degree account for 20%, 30%, and 30% of the total weight respectively. This algorithm is used to accurately calculate the characteristic values ​​of severe weather. , and then judged by the classifier, accurately identify the current blizzard weather.

[0030] The intelligent control and feedback module quickly receives the blizzard weather recognition results, and it uses the intelligent decision fusion algorithm , combined with the real-time feedback of the in-vehicle light data from the light sensor and the feedback data given by the passengers through the feedback terminal, in-depth analysis and intelligent decision-making are carried out to generate accurate light adjustment instructions.

[0031] The light dynamic adjustment module uses an adaptive light control algorithm based on the instructions issued by the intelligent control and feedback module. Finely adjust the light inside the car, including: and The fuzzy logic reasoning system determines the lighting conditions according to the snowstorm weather, ambient light intensity and the preset lighting standards in the car (brightness 400-600lux, color temperature 4000K-6000K, contrast 50%-70%). For reading lights, considering the passengers' needs to read in snowstorm weather, the brightness adjustment range is strictly controlled within ±10% to avoid abrupt changes in light that affect the reading experience. The color temperature is adjusted to 4500K-5500K to simulate natural warm light and make the reading environment more comfortable. In order to ensure the safety of passengers walking in the aisle, the brightness of the aisle lights is increased by 20% and the contrast is increased by 15%, making aisle signs and potential obstacles more clearly visible. The ceiling lights in the car play an overall coordinating role and are adjusted in such a way that the brightness is 1.2 times the average brightness of the reading lights and the aisle lights, the color temperature is the average color temperature of the two, and the contrast is the average contrast of the two, creating an overall lighting environment in the car that is both warm and bright.

[0032] As the train continued to run smoothly, it gradually passed through the mountainous area and entered the plain area. However, the fog gradually became thicker and visibility began to decrease. The camera quickly captured the changes in the surrounding environment again and collected new image information. The severe weather recognition module re-applied the algorithm to calculate the severe weather feature value. After careful analysis, dense fog weather is accurately identified. The intelligent control and feedback module and the light dynamic adjustment module then work closely together to repeat the above process and readjust the light according to the characteristics of dense fog weather.

[0033] During the entire process, the environmental information fusion module deeply integrates the severe weather information with the geographical location of the EMU (mountainous area, plain), operating speed and other information. If the EMU is running at a low speed (0-60km / h), when there is a snowstorm in the mountainous area, considering the complex terrain of the mountainous area, the driver needs a clearer field of vision. The dynamic light adjustment module will increase the light brightness by 30% and the contrast by 20%. When in thick fog on the plains and driving at high speed (161-350km / h), in order to avoid the visual impact of light changes on the driver at high speed, the color temperature will be reduced by 500K and the brightness will be reduced by 10%. At the same time, the uniformity of the light will be enhanced to ensure that the driver can maintain good visual state in different scenarios.

[0034] At the same time, the intelligent control and feedback module has local fault diagnosis and emergency processing functions. It monitors the operating status of each module. If the severe weather characteristic value output by the severe weather recognition module is If the light parameter actually adjusted by the dynamic light adjustment module deviates from the instruction requirement by more than 10%, the fault diagnosis program will be started immediately. If the severe weather recognition module fails, the system will quickly switch to the backup simple weather judgment mode based on the ambient light sensor. When the ambient light intensity change rate is greater than 0.5lux / s, it is judged as severe weather, which is used as a temporary basis for weather judgment. If the dynamic light adjustment module fails, in order to ensure the basic lighting needs in the car, the interior lights will be quickly switched to the preset general lighting mode, with the brightness set to 500lux, the color temperature set to 5000K, and the contrast set to 60%. In addition, when the severe weather characteristic value When the intensity exceeds 1.5 times the upper limit of the normal severe weather range, the system determines that extremely severe weather may occur. At this time, the intelligent control and feedback module will quickly send a warning signal to the central control system of the EMU to remind relevant personnel to make preparations. At the same time, the dynamic light adjustment module will immediately increase the brightness by 30% and the contrast by 25% to cope with the more complex and severe visual environment challenges in extremely severe weather.

[0035] Effects brought about by this embodiment: This embodiment accurately identifies severe weather and adjusts the interior lighting in real time and intelligently according to different weather and environments, thereby creating a comfortable riding lighting environment for passengers and reducing passengers' visual fatigue in different severe weather conditions. At the same time, it provides the driver with a clear field of vision and ensures driving safety. The fusion of environmental information makes the light adjustment more in line with the actual operating scenario and enhances the system's adaptability. The fault diagnosis and early warning mechanism improves the stability and reliability of the system, ensuring that the automatic light control system can operate stably under various complex situations, thereby comprehensively improving the operating quality of the EMU.

[0036] Embodiment 2

[0037] Suppose on a summer afternoon, a high-speed train is traveling between southern cities, passing through large bodies of water and suburban areas, and the weather forecast shows that there will be heavy rain on some sections of the road.

[0038] When the train is driving and approaching the suburbs of the city, the sky gradually darkens and is covered with dark clouds. The high-definition cameras installed on the front, both sides and rear of the train start working in all directions according to the established layout. The three cameras distributed in a fan shape on the front of the train closely capture the field of view directly in front, the cameras on both sides of the train tilted 30 degrees pay close attention to the side environment, and the camera at the rear keeps a close eye on the dynamics behind, collecting images of the surrounding environment in real time.

[0039] The image is transmitted to the severe weather recognition module, which uses the severe weather feature extraction algorithm For rainstorm weather, the weights of raindrop size, speed, and distribution area are 2096, 2096, and 20% respectively, and the remaining 2096 weights are allocated to other related feature areas to accurately calculate the severe weather feature values. , and the classifier identifies the rainstorm weather.

[0040] After receiving the rainstorm weather identification results, the intelligent control and feedback module uses the intelligent decision fusion algorithm , combined with light sensor data and passenger feedback data, quickly analyze and generate light adjustment instructions.

[0041] The light dynamic adjustment module uses the adaptive light control algorithm according to the instructions. Adjust the light inside the car, including: and The fuzzy logic reasoning system controls the reading light brightness adjustment within ±10% and the color temperature is adjusted to 5000K-5500K, which is closer to natural light, according to heavy rain weather, ambient light intensity and preset lighting standards in the car. The aisle light brightness is increased by 25% and the contrast is increased by 20%, allowing passengers to clearly see the surrounding situation when walking in the aisle. The ceiling light of the car follows the established rules and balances its own brightness, color temperature and contrast based on the adjustment parameters of the reading light and aisle light to create a comfortable and unified car lighting environment.

[0042] As the train continued to move forward and entered a section of road near a large body of water, the rainstorm became more severe. At this time, the environmental information fusion module integrated the severe weather information with the train's geographical location (near the water) and operating speed (medium speed 100-160km / h). Taking into account the dual effects of water surface reflection and rainstorm on vision, the intelligent control and feedback module issued special instructions to the dynamic light adjustment module to further enhance the contrast of the light in the car and increase the brightness by 10%, so as to reduce the interference of complex external light on the driver and passengers' vision.

[0043] During the entire operation process, the intelligent control and feedback module continuously monitors the system status. If it detects the severe weather characteristic value output by the severe weather recognition module, If the module fails, the vehicle interior lights will be switched to the preset general lighting mode with a brightness of 500 lux, a color temperature of 5000K, and a contrast ratio of 60%. When the brightness exceeds the upper limit of the normal range by 1.5 times, it is determined that extreme rainstorm may occur. The intelligent control and feedback module sends a warning signal to the central control system of the EMU. At the same time, the dynamic light adjustment module increases the brightness by another 20% and the contrast by 15% to ensure that the interior lighting meets the needs of special weather conditions.

[0044] Effects brought about by this embodiment: In the scenario of this embodiment, the system effectively copes with the complex light changes in rainstorm weather with accurate weather recognition and intelligent light control, provides passengers with a comfortable lighting environment, avoids visual discomfort caused by poor light, and provides drivers with a clear field of vision in complex situations such as rainstorms and water reflections to ensure driving safety. The fault diagnosis and early warning mechanism ensures the stable operation of the system under various emergency situations, comprehensively improving the operational safety and comfort of the train in special weather and complex sections.

[0045] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the same elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A computer vision-assisted automatic light control system for electric vehicles, characterized in that: The system includes: a severe weather recognition module, a light dynamic adjustment module, and an intelligent control and feedback module; The severe weather recognition module is set outside the train and consists of high-definition cameras located at the front, both sides of the train body and the rear of the train. It is used to collect images of the train's surrounding environment in real time and uses a severe weather feature extraction algorithm. The calculation formula is: ,in Represents the severe weather characteristic value, is the corresponding image area The weight coefficient of As the feature enhancement factor, the algorithm is used to extract the key features of the image and the classifier is used to determine the type of severe weather; The light dynamic adjustment module is connected to the severe weather recognition module, and an adaptive light control algorithm is applied based on the recognition result. The algorithm is based on the formula ,in is the adjusted light parameter, is to adjust the front light parameters, is the ambient light intensity, , is the light adjustment factor, , To correct the parameters, the light parameters of the brightness, color temperature and contrast of the interior lights are adjusted accordingly; The intelligent control and feedback module is connected to the severe weather recognition module and the light dynamic adjustment module, receives the severe weather recognition result and the data of the in-vehicle light sensor and the passenger feedback terminal, and uses the intelligent decision fusion algorithm to obtain the formula: ,in For decision-making instructions, is the decision factor weight, Based on the severe weather identification results , light sensor data , Passenger feedback data The decision function is used to optimize and control the light adjustment strategy.

2. According to claim 1, a computer vision-assisted automatic light control system for electric vehicles is characterized in that: The layout of the high-definition cameras in the severe weather recognition module is as follows: three are installed at the front of the vehicle in a fan-shaped distribution, with a viewing angle covering 180° in the front; two are installed on each side of the vehicle body, tilted 30° respectively, and the viewing angle covers the area where the angle between the side of the vehicle body and the direction of travel of the vehicle is 60°-150°; one is installed at the rear of the vehicle, with a viewing angle covering 120° in the rear.

3. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The weight coefficient in the severe weather feature extraction algorithm Through the analysis of historical image data, it is determined that for heavy rain weather, the image area weight of raindrops accounts for 60% of the total weight, of which the weights of raindrop size, speed, and distribution area are 20%, 20%, and 20% respectively. In heavy snow weather, the image area weights of snowflake shape, falling speed, and coverage degree account for 20%, 30%, and 30% of the total weight respectively. In dense fog weather, the image area weights of fog concentration and distribution range account for 40% and 40% of the total weight respectively, and the remaining 20% ​​of the weight is allocated to other feature areas.

4. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The light adjustment coefficient in the light dynamic adjustment module and The fuzzy logic reasoning system determines the current bad weather type, ambient light intensity and preset light standards in the car. The bad weather types are divided into three categories: heavy rain, heavy snow and dense fog. The ambient light intensity is divided into three intervals: low, medium and high, corresponding to 0-100lux, 101-500lux, 501lux and above. The preset light standards in the car are set to brightness 400-600lux, color temperature 4000K-6000K, and contrast 50%-70%. The fuzzy logic reasoning system calculates these variables through specific fuzzy rules. and The numerical value of .

5. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The adjustment strategies of the dynamic light adjustment module for different types of interior lights are as follows: the brightness of the reading light is adjusted within ±10%, the color temperature is adjusted in the range of 4500K-5500K, the brightness of the aisle light is increased by 20% in bad weather, and the contrast is increased by 15%. The ceiling light of the car adjusts its own light parameters according to the adjustment of the reading light and the aisle light, with the brightness being 1.2 times the average brightness of the reading light and the aisle light, the color temperature being the average color temperature of the two, and the contrast being the average contrast of the two.

6. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The decision factor weights in the intelligent control and feedback module , the neural network algorithm uses the running time and mileage of the train as input variables for self-learning and updating. A three-layer fully connected neural network is used, with the input layer being the running time and mileage, and the output layer being the decision factor weights. , with each running 1000 kilometers and 100 hours as an update cycle, calculate the mean square error between the output data of each module and the preset ideal data under different decision instructions, and update it according to the gradient descent method , so that the system can adapt to the actual operation changes of the train and optimize the light adjustment strategy.

7. The automatic light control system for electric vehicles based on computer vision assistance according to claim 1 is characterized in that: The intelligent control and feedback module has fault diagnosis and emergency processing functions. When abnormalities are detected in the severe weather recognition module, the light dynamic adjustment module and the internal data transmission of the system, the intelligent control and feedback module starts the fault diagnosis program. The program quickly locates the fault point by analyzing the rationality of the output data of each module and comparing it with the preset normal operating parameter range, and starts corresponding emergency processing measures for different fault types.

8. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The system also includes an environmental information fusion module, which integrates the weather information obtained by the severe weather identification module with the running speed and geographical location information of the train. The running speed is divided into three intervals: low speed, medium speed and high speed. The geographical location information is obtained by the on-board GPS, including longitude and latitude, altitude and terrain type. According to the combination of weather type, running speed interval and terrain type, control instructions for the dynamic light adjustment module are generated according to the following rules: when it is raining heavily in mountainous areas and driving at low speed, the light brightness is increased by 30% and the contrast is increased by 20%; when it is snowing heavily in plains and driving at high speed, the color temperature is reduced by 500K and the brightness is reduced by 10%, so that the light adjustment fits the actual operating scenario.

9. The computer vision-assisted automatic light control system for electric vehicles according to claim 1, characterized in that: The energy-saving optimization strategy of the dynamic light adjustment module is: when the ambient light intensity change rate is less than 0.2lux / s due to severe weather changes, the light adjustment range does not exceed ±5%, and the adjustment frequency is not higher than once every 5 minutes. The distribution of people is detected by infrared sensors installed on the top of the car and on both sides of the aisle, and the activities of people are judged by the principle of thermal imaging. When there is no human activity in a certain area for 3 consecutive minutes, the light brightness in the area is reduced by 30%. If there is no human activity for 10 consecutive minutes, the lights in the area are turned off, thereby meeting the light needs in the car while reducing the energy consumption of the system.

10. The automatic light control system for electric vehicles based on computer vision assistance according to claim 1, characterized in that: The severe weather recognition module has an early warning function for extremely severe weather. The early warning function is implemented as follows: the upper limit of the normal severe weather range is the 95% quantile of the W value in the historical data, and the severe weather characteristic value is When the intensity exceeds 1.5 times the upper limit of the normal severe weather range, it is determined that extremely severe weather may occur. The severe weather recognition module sends a warning signal to the EMU central control system and sends a special instruction to the light dynamic adjustment module. The light dynamic adjustment module increases the brightness by 30% and the contrast by 25% to cope with the visual environment in extremely severe weather.

Images