Car lamp and intelligent control system

Abstract

The invention relates to the technical field of image processing, in particular to a vehicle lamp and an intelligent control system, and the system comprises the following steps: an environment sensing module which collects a surrounding environment image through a vehicle-mounted camera, carries out the resolution optimization of the image, and generates an optimized image; and based on the optimized image, using deep learning analysis to identify a target environment sign, including a tunnel and a school area, and generating an environment sign identification result. According to the method, on the basis of adjusting the vehicle lamp, the illumination mode is further finely adjusted in combination with the real-time vehicle speed and the weather change condition, the integrating degree of illumination and the real-time driving requirement is improved, and it is ensured that the optimal illumination effect can be obtained in different driving environments; by monitoring the light adjusting effect and the feedback condition of the driver in real time, the illumination performance data are collected in time, dynamic self-adaptive adjustment of the intelligent control system of the vehicle lamp is achieved in a continuous feedback optimization mode, and the stability and safety of the illumination performance are continuously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly to a vehicle headlight and an intelligent control system. Background Art

[0002] Image processing technology refers to the technology of using a computer to analyze, enhance, recognize, and understand the collected image information, mainly involving steps such as image acquisition, preprocessing, feature extraction, pattern recognition, target detection, and classification. In the prior art, there is a lack of effective combination of real-time dynamic factors such as vehicle speed and weather conditions, and it is difficult to perform real-time fine-tuning on the headlight illumination mode, resulting in the actual illumination effect being difficult to continuously meet the needs of drivers and environmental changes. Therefore, improvements are needed. Summary of the Invention

[0003] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose a vehicle headlight and an intelligent control system.

[0004] To achieve the above purpose, the present invention adopts the following technical solutions. An intelligent control system includes the following steps:

[0005] An environment perception module, which collects surrounding environment images through an in-vehicle camera, performs resolution optimization processing on the images to generate optimized images; based on the optimized images, uses deep learning to analyze and identify target environment signs, including tunnels and school areas, to generate environment sign recognition results;

[0006] A light adjustment module, which adjusts the brightness of the vehicle headlights according to the environment sign recognition results, calculates the required light parameters, and generates preliminary light adjustment parameters; based on the preliminary light adjustment parameters, continues to optimize the light projection angle according to the road conditions to generate final light adjustment parameters;

[0007] A lighting mode selection module, which fine-tunes the lighting mode based on the final light adjustment parameters, combines the current vehicle speed and weather conditions, and generates an optimized lighting mode result;

[0008] A feedback optimization module, which uses the optimized lighting mode result to monitor the lighting effect and driver feedback in real time, collects lighting performance data, and generates lighting performance feedback data.

[0009] Preferably, the step of obtaining the optimized image is as follows:

[0010] The in-vehicle camera is started, and a continuous image sequence of the surrounding environment is collected, and color balance and brightness adjustment are performed to obtain a processed environment image;

[0011] According to the processed environment image, a high-pass filter is applied to remove noise in the image, and edge detection is used to enhance the object edges in the image to obtain an environment image with optimized resolution;

[0012] Based on the environment image optimized according to the resolution, the features in the image are refined through image sharpening to generate an optimized image.

[0013] Preferably, the steps for obtaining the environmental sign recognition result are as follows:

[0014] Based on the optimized image, image features are extracted, and roads, buildings, and traffic signs are separated from the background area through image segmentation to obtain candidate areas for environmental signs;

[0015] According to the candidate areas for environmental signs, a deep learning model is applied for classification, input feature vectors are used for target matching, the matching degree is calculated, and environmental signs are screened to generate an environmental sign classification result;

[0016] Based on the environmental sign classification result, confidence evaluation is performed, misrecognized targets are screened out, the positions and categories of target environmental signs are determined, and an environmental sign recognition result is generated.

[0017] Preferably, the steps for obtaining the preliminary light adjustment parameters are as follows:

[0018] Based on the environmental sign recognition result, the positional relationship between the recognized environmental sign type and the current position of the vehicle is determined, the real-time distance between the vehicle and the environmental sign is calculated, and light adjustment requirement data is obtained;

[0019] According to the light adjustment requirement data, the required headlight brightness adjustment number is calculated, and the calculation formula is:

[0020]

[0021] Wherein, represents the required headlight brightness adjustment number, represents the highest set value of the headlight brightness, represents the lowest set value of the headlight brightness, represents the real-time measured distance between the vehicle and the environmental sign, represents the safety reference distance, represents the height difference value between the road where the current vehicle is traveling and the environmental sign, represents the initial angle of the current headlight light projection;

[0022] Based on the required headlight brightness adjustment number, real-time brightness synchronization adjustment of the headlights is performed to obtain preliminary light adjustment parameters.

[0023] Preferably, the steps for obtaining the final light adjustment parameters are as follows:

[0024] Based on the preliminary light adjustment parameters, detect the light reflectivity of the road surface, analyze the reflection characteristics of the light, calculate the initial scattering range of the light projection, and obtain the light projection adjustment data;

[0025] According to the light projection adjustment data, calculate the optimized light projection angle, and the calculation formula is:

[0026]

[0027] Wherein, represents the optimized light projection angle, represents the height of the vehicle headlight position relative to the road, represents the height of the target area, represents the horizontal distance between the vehicle and the target area, represents the light reflectivity of the road surface, represents the standard light reflectivity;

[0028] Based on the optimized light projection angle, adjust the angle of the vehicle headlight beam to generate the final light adjustment parameters.

[0029] Preferably, the steps for obtaining the optimization result of the lighting mode are:

[0030] Based on the final light adjustment parameters, obtain the real-time speed information of the current vehicle, detect the road visibility and the surrounding ambient light, and obtain the lighting mode adjustment data;

[0031] According to the lighting mode adjustment data, calculate the lighting mode optimization index, and the calculation formula is:

[0032]

[0033] Wherein, represents the lighting mode optimization index, represents the current driving speed of the vehicle, represents the light influence factor, represents the detected light intensity in front of the current environment, represents the ambient light intensity around the vehicle, represents the detected distance between the vehicle and the obstacle in front;

[0034] Based on the lighting mode optimization index, select the vehicle headlight adjustment strategy to obtain the optimization result of the lighting mode.

[0035] Preferably, the steps for obtaining the lighting performance feedback data are:

[0036] Based on the optimization result of the lighting mode, obtain the brightness distribution of the current vehicle headlight illumination area, and combine the road reflection situation obtained by the camera in front of the vehicle to obtain the lighting monitoring data;

[0037] Calculate the lighting performance evaluation value according to the lighting monitoring data, and the calculation formula is:

[0038]

[0039] where, represents the lighting performance evaluation value, represents the average brightness of the headlight irradiation area, represents the reflection coefficient of the road surface, represents the background brightness from the driver's perspective, represents the monitoring time interval, represents the average driving speed of the driver during the monitoring time, represents the recommended driving speed in the current environment;

[0040] Based on the lighting performance evaluation value, analyze the deviation between the driver's manual adjustment behavior and the system adjustment, extract the effective lighting adjustment trend, and generate lighting performance feedback data.

[0041] The present invention also provides a headlight applicable to the intelligent control system in any one of the above, including: a heat dissipation lamp housing with ventilation openings on the outer wall, a heat dissipation fan corresponding to the ventilation openings is installed inside the heat dissipation lamp housing, a conductive component for power supply is clamped at one end of the heat dissipation lamp housing, a copper substrate is arranged inside the heat dissipation lamp housing, LED light source chips are arranged on both sides of the copper substrate, a connection terminal for electrically connecting two LED light source chips is arranged at one end of the copper substrate, a bifurcated part adapted to the heat dissipation fan is arranged at the other end of the copper substrate, an electronic control chip is arranged on one side of the copper substrate, heat conduction pipe fittings are arranged on both outer sides of the copper substrate, and the heat dissipation fan is located between the electronic control chip and the conductive component;

[0042] A cavity for accommodating the heat dissipation fan and the conductive component is formed inside the heat dissipation lamp housing, and an opening corresponding to the light emitting surface of the LED light source chip is formed outside the heat dissipation lamp housing, and the opening is arranged outside near the ventilation opening.

[0043] Further, the conductive component includes a conductive plug and a rectifying plate member, and card slots a for welding the pins of the conductive plug and card slots b for welding the bifurcated part of the copper substrate are respectively arranged around the rectifying plate member.

[0044] Further, a heat dissipation round hole for the end part of the heat conduction pipe fitting to extend out is formed at the other end of the heat dissipation lamp housing, and the end parts of two heat conduction pipe fittings are aligned with the bifurcated part of the copper substrate.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] In the present invention, through deep learning, the target environmental signs are accurately recognized, the required light parameters are automatically calculated, and the brightness and irradiation angle of the vehicle lamp are finely adjusted, so that the light output by the vehicle lamp is accurately matched with the actual road conditions, solving the problem of blurred vision caused by too high or too low light intensity, and ensuring a clear and comfortable driving vision; on the basis of adjusting the vehicle lamp, the lighting mode is further finely adjusted in combination with the real-time vehicle speed and weather change conditions, improving the matching degree of the lighting and the real-time driving requirements, and ensuring the best lighting effect in different driving environments; by monitoring the light adjustment effect and the driver's feedback in real time, the lighting performance data is collected in time, and the dynamic adaptive adjustment of the intelligent control system of the vehicle lamp is realized in a continuous feedback optimization manner, continuously improving the stability and safety of the lighting performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of the steps of the present invention;

[0048] Figure 2 is a schematic diagram of the overall external structure of the vehicle lamp in the present invention;

[0049] Figure 3 is a schematic diagram of the overall exploded structure of the vehicle lamp in the present invention;

[0050] Figure 4 is a schematic diagram of the overall external structure of the vehicle lamp in another perspective in the present invention.

[0051] Figure 5 In the present invention Figure 3 is an enlarged schematic diagram of part A.

[0052] Reference numerals:

[0053] 1. Heat dissipation lamp housing; 10. Ventilation opening; 11. Heat dissipation fan; 12. Opening; 13. Heat dissipation round hole; 2. Conductive component; 21. Conductive plug; 22. Rectifying plate member; 221. Slot a; 222. Slot b; 3. Copper substrate; 30. LED light source chip; 301. Connection terminal; 31. Electric control chip; 4. Heat conduction pipe fitting. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0055] Please refer to Figure 1 , the present invention provides a technical solution, an intelligent control system, including the following steps:

[0056] The environmental perception module collects images of the surrounding environment through an in-vehicle camera, optimizes the resolution of the images, and generates optimized images; based on the optimized images, it uses deep learning to analyze and identify target environmental signs, including tunnels and school areas, and generates environmental sign recognition results;

[0057] The light adjustment module adjusts the brightness of the vehicle lights according to the environmental sign recognition results, calculates the required light parameters, and generates preliminary light adjustment parameters; based on the preliminary light adjustment parameters, it further optimizes the light projection angle according to the road conditions and generates the final light adjustment parameters;

[0058] The lighting mode selection module fine-tunes the lighting mode based on the final light adjustment parameters, combines the current vehicle speed and weather conditions, and generates an optimized lighting mode result;

[0059] The feedback optimization module uses the optimized lighting mode result to monitor the lighting effect and driver feedback in real time, collects lighting performance data, and generates lighting performance feedback data.

[0060] The steps for obtaining the optimized image are as follows:

[0061] The in-vehicle camera is started, a continuous image sequence of the surrounding environment is collected, color balance and brightness adjustment are performed, and a processed environmental image is obtained;

[0062] Based on the processed environmental image, a high-pass filter is applied to remove noise in the image, and edge detection is used to enhance the object edges in the image, obtaining an environmental image with optimized resolution;

[0063] Based on the environmental image with optimized resolution, the features in the image are refined through image sharpening to generate the optimized image.

[0064] Specifically, based on the obtained continuous image sequence, first read the color distribution of each frame of the image and count the red, green, and blue channel values of each pixel one by one. If the statistical result is outside the range of 0 to 255, it is regarded as an outlier and excluded, thus forming preliminary pixel color data. Subsequently, based on this pixel color data, calculate the average value and variance of each channel, and refer to the standard color balance range set according to previous experience. For example, set the balance range of the red channel between to , set the balance range of the green channel between to , set the balance range of the blue channel between to . Determine whether to perform channel enhancement or attenuation by comparing the difference between the current pixel distribution and this balance range, and further adjust the overall brightness. Set a brightness reference value, for example , calculate the brightness mean of each frame And subtract from If the result of subtraction is greater than the deviation threshold set by experience , then perform linear correction on the brightness of the entire frame of the image. For example, let , and then subtract the brightness value of each pixel by to achieve equalization processing. At the same time, specifically limit and truncate extremely over-bright or over-dark areas. If some pixel values still exceed the allowable range after correction, directly clip these pixel values to 0 or 255. Finally, perform another comparison on all frames uniformly. After confirming that the aforementioned color balance and brightness range are satisfied, obtain the processed environmental image.

[0065] According to the processed environmental image, first perform a preliminary scan on the overall picture and record the positions of noise points. To determine the noise threshold, several random regions can be selected in the entire image and the gray variance can be measured. If the measured variance is higher than the upper limit value set by previous manual experience, it can be determined that there are obvious noise points in this region. Then, use the high-pass filtering method to remove the noise points in these regions. When selecting the high-pass filtering kernel, a central protruding kernel with a size of 3×3 or 5×5 can be used. By calculating the local gray difference, the details of the image are retained while the noise is reduced as much as possible. Compare the difference between each pixel after filtering and its surrounding pixels with an empirically obtained noise rejection threshold. If the absolute value of the difference is greater than, for example a certain value, then reassign the gray level of this pixel. Next, perform edge detection on the image after noise rejection. The specific method is to traverse the horizontal and vertical neighborhoods of each pixel in the image. When using the Sobel or Canny operator, it is necessary to determine the convolution mask and the non-maximum suppression threshold. For example, divide the gradient intensity into the range of 0 to 255. Gradient values below 30 are not marked, and gradient values above 30 are marked as potential edges. These potential edges are screened according to whether they exceed the high threshold of 40 to obtain the final edge pixels. After completing the edge detection, obtain the initially enhanced contour information, and merge and compare it with the high-pass filtering result of the previous step. If some local details are lost due to incorrect threshold setting, the threshold range can be fine-tuned again, and finally obtain the environmental image with optimized resolution.

[0066] Based on the environmental image with optimized resolution, read the distribution information of each pixel in the spatial domain and the frequency domain. When sharpening the image, the Laplacian operator can be used to further magnify local details. When setting the Laplacian operator, a central negative coefficient template of 3×3 or 5×5 is usually considered, and a threshold such as is defined to determine the sharpening intensity. After calculating the Laplacian response, multiply it by Add back the original image pixel values. If some pixels exceed 255 after adding back, truncate them to 255, and if they are below 0, truncate them to 0. At this time, this operation can be performed independently for each channel to avoid color cast problems caused by multi-channel mixing. For high-contrast regions that appear in the image, they can be cropped according to the highlight threshold and shadow threshold obtained from previous tests. For example, the highlight threshold is 200 and the shadow threshold is 30. If local pixels exceed these two ranges after sharpening, then perform equalization processing. At the same time, compare the mean square error of the possible noise again. If the mean square error after sharpening is greater than 15, it means that the noise has been amplified too much and the threshold needs to be reduced. Recalculate. After each region of the image meets the previously established brightness and sharpness ranges, generate an optimized image.

[0067] The steps for obtaining the environmental sign recognition result are as follows:

[0068] Based on the optimized image, extract image features, separate roads, buildings, and traffic signs from the background area through image segmentation to obtain environmental sign candidate regions;

[0069] According to the environmental sign candidate regions, apply a deep learning model for classification, input feature vectors for target matching, calculate the matching degree, and screen environmental signs to generate an environmental sign classification result;

[0070] Based on the environmental sign classification result, perform confidence evaluation, screen out misrecognized targets, determine the position and category of the target environmental signs, and generate an environmental sign recognition result.

[0071] Specifically, based on the optimized image obtained previously, first read and record the brightness and color information of each pixel in the RGB space, and then select the region-growing method to aggregate adjacent pixels with similar features in the image. To set the aggregation conditions, the distribution ranges of different targets such as roads, buildings, and traffic signs in terms of color and texture can be extracted from actual observation experience. For example, the gray levels of road pixels are concentrated in the to interval and the texture gradient is lower than , the gray levels of building pixels are concentrated in the to interval and the texture gradient is approximately between to The color distribution of traffic signs is such that the red or blue channel value is higher than Mainly supplemented by obvious edge gradients, the similarity with known categories can be judged pixel by pixel according to these ranges and continuously expanded and aggregated around. If the color and texture of the current pixel both fall within the range of the same category, it is classified into the area of that category; otherwise, continue to compare with other categories or mark it as an area without attribution. To make the segmentation more accurate, threshold segmentation can be used for secondary refinement after region growing. The setting of the threshold is calculated from the previous data. For example, the color threshold is determined as and the texture threshold is determined as . If the feature difference of a certain pixel from the target category exceeds or , it will no longer be incorporated into the area of that category. After the pixel-level division of the entire image is completed, road pixels, building pixels, and traffic sign pixels are separated from the remaining background areas. When summarizing these separated connected areas, small scattered noise areas with too small areas can be further removed. The minimum connected domain area pixels is set according to experience. When the area of a certain connected area is less than , the area is directly discarded. After the above operations, the candidate areas of environmental signs are obtained.

[0072] According to the candidate areas of environmental signs obtained in the previous step, first, the pixel feature vectors in each candidate area are constructed into input data that can be recognized by the deep learning model. The specific method is to obtain parameters such as contour shape, typical texture, and average brightness in each collected candidate area, and at the same time record the position of the area in the original image coordinates. To train the deep learning model, a batch of labeled environmental sign image samples need to be prepared and maintain diversity in the training set and the validation set. For example, collect traffic sign images in the actual road environment sheets and include common buildings or road backgrounds as negative samples . These samples are input into a convolutional neural network containing a multi-layer convolutional and pooling structure for offline training. The learning rate can be set to and the random gradient descent method is used to continuously update the network weights in rounds of iterations. After the loss function converges, the model parameters obtained from the training are solidified. During inference, the feature vectors of the aforementioned candidate areas are sent into the network and the matching probabilities of each category are output. Whether it is a traffic sign category is screened by comparing with a pre-set classification threshold. The classification threshold can be determined by the comprehensive performance of the F1 score on the validation set. For example, take as the determination threshold. If the probability of a certain area for the traffic sign category in the network output is greater than

[0073] , the area is retained and used as the target sign. At the same time, the same judgment is made for the building and road categories and the unmatched areas are removed. Finally, the corresponding prediction labels are marked on each retained area and the labels and candidate area coordinate information are summarized together to generate the environmental sign classification result.Based on the above environmental sign classification results, first traverse all regions predicted as traffic signs or buildings and obtain their model output probabilities. Subsequently, perform confidence evaluation for each region. To set the confidence threshold, the probability interval with a relatively high model prediction accuracy in the training set can be statistically obtained based on the previous annotation data. For example, when the probability is higher than there are very few false predictions. If the predicted probability of a certain region is lower than it is directly determined as a misidentification and discarded. For the results between and shape feature comparison is supplemented and the difference between its actual area and the shape of typical traffic signs or buildings is observed. If the difference is within and the texture distribution difference is not greater than the region is retained; otherwise, it is excluded. Through such multiple determinations, misidentified targets are further excluded and regions with higher confidence are retained. Finally, record the image coordinate positions, corresponding category labels, and matching probabilities of the retained regions. After summarizing all the region information that meets the confidence requirements, the environmental sign recognition result is generated.

[0074] The steps for obtaining the preliminary light adjustment parameters are as follows:

[0075] Based on the environmental sign recognition result, determine the positional relationship between the recognized environmental sign type and the current position of the vehicle, calculate the real-time distance between the vehicle and the environmental sign, and obtain the light adjustment requirement data;

[0076] According to the light adjustment requirement data, calculate the required headlight brightness adjustment number. The calculation formula is:

[0077]

[0078] Among them, represents the required headlight brightness adjustment number, represents the highest set value of the headlight brightness, represents the lowest set value of the headlight brightness, represents the real-time measured distance between the vehicle and the environmental sign, represents the safety reference distance, represents the height difference value between the road where the current vehicle is traveling and the environmental sign, represents the initial angle of the current vehicle's headlight light projection;

[0079] Based on the required headlight brightness adjustment number, perform real-time brightness synchronization adjustment of the headlights to obtain the preliminary light adjustment parameters.

[0080] Specifically, based on the sign category and coordinate reference information provided by the environmental sign recognition result, read the coordinate positioning data of the vehicle's current driving route and compare the coordinate positions of the environmental signs item by item. Calculate the horizontal difference and elevation difference between the two in the plane coordinate system. Obtain the vehicle's current actual position by collecting the longitude and latitude data during the vehicle's driving and matching it with the pre-established road geographical information. Combine the geographical positions of the environmental signs marked in the map information and calculate the horizontal and vertical distances between the two in pairs. Compare these distance values with the pre-established confidence intervals respectively. For example, consider the range from 0 meters to 200 meters as the short-distance range, 201 meters to 1000 meters as the medium-distance range, and more than 1000 meters as the long-distance range. If a certain calculation result significantly exceeds the confidence interval, record this value for review and re-correct the position data after confirming the true information of the road where the vehicle is located. Subsequently, expand the straight-line distance between the vehicle and the environmental sign using the Pythagorean theorem. If the elevation difference is denoted as Zh and the planar distance is denoted as Lw, the real-time distance can be calculated and converted into the final measurement value. In actual measurement, if the elevation difference is less than 2 meters, it can be simply ignored. Otherwise, the vertical accuracy must be ensured to be within 0.5 meters. Then, compare the interval change of this distance value in combination with the pre-determined safety reference standard. If the distance is less than 20 meters, classify it into the short-distance interval and mark it as needing further attention. If the distance is greater than 500 meters, mark it as the long-distance interval and no special treatment is required. Finally, create and summarize the light adjustment requirement data based on these screened and corrected distance data to obtain the light adjustment requirement data.

[0081] The advantage of the formula is that by introducing multiple parameters such as the actual distance between the vehicle and the environmental sign, the road height difference, and the initial angle of the headlight projection, the refined calculation of the headlight brightness is realized, so as to obtain more flexible brightness adjustment numbers under different road conditions and environmental sign types.

[0082] The steps for obtaining the parameters are as follows: Before the vehicle lighting system leaves the factory, calibrate the maximum power output of the headlight, count the corresponding maximum brightness value, and measure this maximum brightness value several times during the vehicle installation and commissioning. Incorporate the maximum brightness values under different measurement conditions into a data sequence , and then calculate the mean value of this sequence as the reference value. If it is found in subsequent inspections that the maximum brightness deviates due to extreme environments, additional measurement values are supplemented and the mean value is updated. See the following formula: , where Denote the number of measurements. Each measurement is completed in a darkroom without external light interference. The calibration error of the photometer needs to be controlled within 5% in advance. For example, for a certain model of vehicle headlight, ten values 4500, 4580, 4420, 4550, 4510, 4490, 4570, 4600, 4520, and 4480 are obtained in ten measurements, then the calculated result is .

[0083] The steps to obtain the parameter are as follows: Set the lowest perceivable brightness level in the same vehicle lighting system, record the actual measurement data of the vehicle headlight in the minimum safety brightness state, and unify the values obtained from each measurement into a data sequence . According to the same mean value processing, is obtained. For example, when a number of measured low brightness values are concentrated in the range of 500 to 600, can be comprehensively obtained.

[0084] The steps to obtain the parameter are as follows: Based on the real-time distance between the vehicle and the environmental marker recorded in the light adjustment requirement data obtained previously, directly read a numerical sequence that has completed measurement and calibration, and denote this sequence as . During each vehicle driving period, a large amount of ranging information is collected, and then this information is tested under the condition of unobstructed straight-line line-of-sight. For example, when the vehicle is driving in a certain section, if the average value of the ranging results is 30 meters and the error is within 0.5 meters, then .

[0085] The steps to obtain the parameter are as follows: Refer to the previously determined safety reference distance data set. This distance is calibrated in advance according to the road type and speed limit range. Determine the safety reference distance of each road through continuous months of road driving monitoring statistics and safety distance determination criteria, and record the corresponding for the current road. Select the corresponding safety reference distance value for use. For example, on an urban road with a designed vehicle speed of 60 kilometers per hour, the most common safety reference distance obtained through statistics is 50 meters. At this time, take .

[0086] The steps to obtain the parameter are as follows: When the vehicle is driving on different roads, record the height difference between the road location and the base or installation point of the environmental marker through the elevation surveying system. If this height difference sequence is , then select the item closest to the current driving position as value. For example, when the vehicle is driving on an elevated section and is 2 meters different from the ground column where the marker is located, then directly take .

[0087] The steps for obtaining the parameters are as follows: Using the initial calibration value of the headlight installation angle and performing actual measurements when the vehicle enters the workshop for maintenance. For example, record the angle between the headlight projection beam and the ground horizontal line through a laser angle measuring instrument, and count the final angle values under different installation error conditions. Calculate the average of this sequence as the initial angle of the headlight beam projection. For example, in one measurement, if the five data items are 3.5 degrees, 3.4 degrees, 3.6 degrees, 3.5 degrees, and 3.5 degrees, then .

[0088] Calculation process:

[0089] First step, read the required parameters , for example, take , , , , , ;

[0090] Second step, calculate , that is ;

[0091] Third step, calculate the denominator , where , , so the denominator is approximately ;

[0092] Fourth step, calculate the numerator ;

[0093] Fifth step, first find the first item , then add the second item , and combine to get ;

[0094] The result shows that under the current environment and parameters, the headlight brightness adjustment number is approximately 96981.57. If there are more refined brightness limits or safety limit requirements in subsequent judgments, this value can be further trimmed or corrected. Different numerical ranges represent different degrees of brightness output requirements. For example, if the final value is greater than 50000, it means that the brightness needs to be significantly increased for long distances or large height differences, otherwise, when the value is less than 5000, the brightness output level can be maintained at a lower level.

[0095] Based on the required headlight brightness adjustment values, traverse the data recorded for each time segment during the current vehicle driving process in sequence. According to the vehicle speed and turning angle, refer to the pre-established brightness adjustment table. If the vehicle is on a straight section and the speed is greater than 40 kilometers per hour, perform fast brightness synchronization according to the previously obtained brightness adjustment values. Otherwise, in a curve scenario where the turning radius is less than 30 meters, first compare the lateral lighting requirements and combine with the road lighting conditions for secondary correction. Set up a lateral correction coefficient to measure the adjustment of the diffusion angle required for the headlights when turning. If this coefficient is denoted as Mt and is obtained by combining the curve radius data with the vehicle driving speed data, in the previously statistically measured value range, when the curve radius is less than 30 meters and the speed exceeds 35 kilometers per hour, Mt can be in the range of 0.3 to 0.4. Then, use the actually measured direction angle to compare with the empirical threshold to determine the final correction amount. For example, consider a direction angle greater than 10 degrees as a large turn and increase the brightness output by about 10%. After recording, if it is found that the vehicle is in an elevated or tunnel area, geographical information identification can be further introduced to slightly increase or decrease the brightness. After comprehensively comparing these parameters, synchronously adjust the headlight output value at each moment, associate the effective points of all adjustment actions with each vehicle state point for subsequent tracking and debugging accuracy. Finally, generate preliminary light adjustment parameters based on the sorted adjustment information.

[0096] The steps to obtain the final light adjustment parameters are as follows:

[0097] Based on the preliminary light adjustment parameters, detect the light reflectivity of the road surface, analyze the reflection characteristics of the light, calculate the initial scattering range of the light projection, and obtain the light projection adjustment data;

[0098] According to the light projection adjustment data, calculate the optimized light projection angle. The calculation formula is:

[0099]

[0100] Wherein, represents the optimized light projection angle, represents the height of the headlight position relative to the road, represents the height of the target area, represents the horizontal distance between the vehicle and the target area, represents the light reflectivity of the road surface, represents the standard light reflectivity;

[0101] Based on the optimized light projection angle, adjust the angle of the headlight beam to generate the final light adjustment parameters.

[0102] Specifically, based on the preliminary light adjustment parameters obtained above, first read the actual road surface material information of the road section where the vehicle is located and record the corresponding reflection data. Then, measure the local reflectivity value of the road surface point by point through a device with a photometric monitoring function. To set the detection accuracy, the range can be divided into several scales between 0 and 1, and the measurement error is required to be less than 0.02. If it is found that some measured values ​​are obviously higher than 0.8 or lower than 0.05, re-measure again. After completing the calibration of several sampling points of the entire road section, the data of these sampling points that exceed or are lower than the established threshold range are summarized and compared for the second time. For example, when the measured road section is an asphalt road surface, the common reflectivity range is between 0.05 and 0.25. If a point is higher than 0.25, it is regarded as a bright section and recorded. If a point is lower than 0.05, it is regarded as an extremely dark section and recorded. These recorded points are compared with the road curvature, terrain height difference and surrounding The information such as the interference degree of the edge light source is correlated and sorted, and then when reading the light reflection characteristics, the reflection behavior of each sampling point under different illumination intensities is compared one by one against the photometric correction formula obtained before. If the reflection coefficient of a certain point shows an increase of more than 0.2, it is marked as a high reflection risk point. If the reflection coefficient of a certain point always fluctuates around 0.06, it is marked as a low reflection point. Then, a scattering range estimation model is constructed based on these marked information. The model will refer to the measurement difference between adjacent points. If it is less than 0.01, it means that the reflectivity of the area is relatively close and can be merged for processing. If the difference is greater than 0.03, it means that the reflectivity of the area is significantly different and needs to be processed in sections. By comparing the average reflectivity of the entire road section and the distribution of peak and valley values, the initial scattering range that may be generated by light projection is determined, and the results are coupled item by item with the preliminary light adjustment parameters obtained previously, and the light projection adjustment data is output.

[0103] The benefit of the formula is that by introducing the parameters of road height, target area height, horizontal distance, road surface reflectivity and standard reflectivity for comprehensive calculation, the light beam angle of the car can be dynamically corrected under different height differences and road surface light reflection conditions.

[0104] The steps for obtaining the parameters are as follows: using the distance and height measuring device at the front end of the vehicle to detect the vertical distance of the headlight position relative to the ground reference plane, first measure the height from the center point of the headlight to the ground multiple times under flat road conditions, and record a set of values , combined with the suspension changes of the vehicle in full and empty states, the calculation formula , select the average value as the final headlight installation height. If the deviation between a single measurement and the average value exceeds 0.02 meters during the test, repeat the measurement. After several measurements, the heights of a certain model of small car were 0.63 meters, 0.62 meters, 0.61 meters, 0.62 meters, and 0.63 meters. The average value is about 0.622 meters and is used as The parameter value of .

[0105] The steps for obtaining the parameter are as follows: for the location of the target area, use the elevation difference between the vertical scale measurement result of the target area itself and the road reference plane as the input data.

[0106] The steps for obtaining the parameter are as follows: when the vehicle is continuing to drive, locate the horizontal distance from the vehicle lamp to the target area, and perform triangulation calculation by combining the navigation positioning data carried by the vehicle and the target area coordinate information. It is also possible to directly obtain several measurement values by using the laser ranging method under the condition of straight-line visibility. , take the average value with reference to the following formula: , if 28 meters, 29 meters, 30 meters, and 29 meters are obtained in four measurements, then .

[0107] The steps for obtaining the parameter are as follows: through the light projection adjustment data and multi-point reflectivity monitoring results obtained previously, search for and match the road surface reflectivity of the target area. If it is found that the area is between 0.15 and 0.25 in the measurement value set, then the numerical interval can be finely statistically analyzed, and then according to the following formula: , perform weighted summation, represents the reflectivity value of the t-th measurement point, represents the weight value related to the lighting angle or light intensity of the measurement point to ensure the consistency of the measurement conditions. If the reflectivities of 6 measurement points collected around the target area are 0.18, 0.20, 0.22, 0.21, 0.19, and 0.20 respectively, and the weights of each point are around 0.95, then the comprehensive calculation can obtain .

[0108] The steps for obtaining the parameter are as follows: taking the reference reflectance value of the conventional asphalt surface of the municipal road under standard test conditions as the benchmark, the statistical department or testing institution generally gives the road reference reflectance with a value in the range of 0.15 to 0.20. After repeatedly measuring this range, select 0.18 as the standard light reflectance. , if it is found in the measurement that the actual extensive average value is exactly 0.18, then is brought into the operation.

[0109] Calculation process:

[0110] The first step is to read the parameters of the current vehicle driving environment: , , , , ;

[0111] The second step is to calculate , substitute the data as , then take (since the angle value is small, the radian is close to it);

[0112] The third step is to calculate , its numerator , denominator , so this term is ;

[0113] The fourth step is to sum up to get , convert the radian system result to an angle of approximately , in order to avoid negative angles, it can be corrected upward or downward based on the vehicle installation reference;

[0114] The result shows that after calculating with the current horizontal distance, height information and road surface reflectivity, the optimized projection angle obtained is approximately , when this value is between -2 degrees and -0.5 degrees, it means the car lights need to be slightly tilted downward, and if it is greater than 0 degrees, it means the car lights need to be adjusted upward. For different value ranges, fine-tuning can be continued in combination with the vehicle state.

[0115] Based on the optimized light projection angle obtained previously, read the speed, turning radius and surrounding light data during the vehicle's driving process every second, compare the above values with the previously established beam dynamic adjustment rules one by one. For example, when the turning radius is less than 40 meters and the speed exceeds 30 kilometers per hour, an additional lateral correction coefficient will be added. If the measured lateral correction coefficient is in the range of 0.2 to 0.3, a corresponding offset value will be added based on the current light projection angle. If the surrounding illuminance measured by the sensor is below 200 lux, an additional downward tilt of 0.5 to 1 degree will be added to the current angle. If the illuminance exceeds 600 lux, only a fine-tuning of 0.1 to 0.3 degrees will be made on the original angle. After summarizing the angle correction process corresponding to each moment, a complete time-series change record of the projection angle is formed, and it is cross-checked with the synchronous data of the vehicle position and speed to check whether there is excessive or insufficient adjustment. To ensure the recording accuracy, the angle changes every 10 seconds can be included in the comparison and the difference interval from the target angle can be marked. For example, if the difference is greater than 1 degree, it will be included in the key inspection sequence. If the difference is within 0.3 degrees, it is considered to meet the normal deviation. Finally, after completing the time-series inspection, the final light adjustment parameters are obtained.

[0116] The steps to obtain the optimized result of the lighting mode are as follows:

[0117] Based on the final light adjustment parameters, obtain the real-time speed information of the current vehicle, detect the road visibility and the surrounding environmental light, and obtain the lighting mode adjustment data;

[0118] According to the lighting mode adjustment data, calculate the lighting mode optimization index, and the calculation formula is:

[0119]

[0120] Among them, represents the lighting mode optimization index, represents the current driving speed of the vehicle, represents the light influence factor, represents the light intensity detected in front of the current environment, represents the ambient light intensity around the vehicle, represents the detection distance between the vehicle and the obstacle in front;

[0121] Based on the lighting mode optimization index, select the headlight adjustment strategy to obtain the lighting mode optimization result.

[0122] Specifically, based on the final parameters of light adjustment, first retrieve the speed data of the vehicle during the current driving process and record it synchronously with the vehicle position frame by frame, compare the speed value with the road speed limit range, for example, when the road speed limit is 60 kilometers per hour, if it is found that the speed data approaches or exceeds 60 kilometers per hour for many times, it is marked as a high speed state, and then read the road visibility data at adjacent times. The visibility data can be measured by collecting information such as image clarity and visible distance through the imaging device installed on the front of the vehicle and combined with the corresponding distance scale. If the visibility is less than the threshold of 300 meters set in advance based on a large amount of field observation experience, it is marked as a low visibility event, and if the visibility is greater than 800 meters, it is marked as a high visibility event. Then link the surrounding environment light data to compare with the speed state and visibility state just now. For example, the light intensity unit is set to lux. It can reach tens of thousands of lux or even 100,000 lux when the sun is directly shining during the day, and it can drop to hundreds of lux or even tens of lux on cloudy days or at night. Select the light measurement results of multiple time periods for comparison to obtain the light fluctuation and record when the light is lower than 100 When the light is less than 10,000 lux, it is classified as a low light state, and when the light is greater than 10,000 lux, it is classified as a high light state. These state marks are combined with the vehicle speed and road visibility for judgment. If the speed is high and the visibility is low and the surrounding light is also dark, a higher demand value is output at the current moment. If the speed is medium and the visibility is in a safe range and the surrounding light is close to the normal value during the day, a lower demand value is output. These demand values ​​are stored in a data sequence in a quantitative manner to represent the vehicle's current adjustment tendency for the lighting mode. The data sequence is paired with the corresponding timestamp of the final light adjustment parameter formed previously, and whether extreme conditions occur in the stock period one by one. If a rapid up and down fluctuation exceeds the 30% range in a short period of time, it is marked as a fluctuation abnormal point, and then the data of all adjacent time points are read for interpolation confirmation and removal of abnormal values ​​that may be caused by instantaneous interference. When most of the data in the overall period is stable or only jitters in a small range, the measurement result is determined to be true and valid. Finally, all the above speed, visibility, and light comprehensive values ​​are summarized to generate lighting mode adjustment data.

[0123] The benefit of the formula is that it incorporates multiple data such as vehicle real-time speed, lighting influence factor, front lighting intensity, ambient light intensity, and obstacle distance into the same expression for calculation, taking into account various objective conditions that may appear on the road, so as to achieve the same indicator. Comprehensive assessment is carried out in

[0124] The parameter acquisition step is to collect driving speed data through the vehicle speed sensor installed on the vehicle. The sensor is calibrated before leaving the factory and dynamically calibrated during the actual road driving of the vehicle. The speed value collected every second is recorded to form a sequence , the average speed representing the current period is obtained by using average processing or median filtering to remove obvious abnormal fluctuations, and then a speed confidence interval is set up in combination with the maximum allowable speed and instantaneous acceleration of the vehicle. When the detected speed value is stable within this interval, it can be used as the operation value. For example, when the average driving speed on an urban road is recorded as 35 kilometers per hour and the deviation does not exceed 2 kilometers per hour after multiple measurements, it can be used for subsequent calculations.

[0125] The steps for obtaining the parameter are as follows: The influence of surrounding light on the vehicle lighting demand is quantified by using a photometric sensor and a field monitoring device. First, the corresponding light level at different lux values needs to be measured in an experimental environment, and then an influence coefficient is assigned to each light level, and a weighted calculation in the following form is used: , where represents the light level bin value of the i-th photometric monitoring, represents the weight value related to the observation angle or light stability. After multiple rounds of collection, 0 to 50 lux can be used as the extremely dark interval and corresponding mapping values with higher influence coefficients, and above 10,000 lux can be used as the bright interval and corresponding lower mapping values. After sorting, if is about 15, it means that the current light is overall in a weak or dark light condition. If is about 3, it indicates that the surrounding light is sufficient and the visibility is good. For example, when the photometric monitoring values are measured between 20 and 60 lux multiple times on a suburban road with few street lights at night, can be finally calculated.

[0126] The steps for obtaining the parameter are as follows: The light intensity in the front of the current environment is detected. Usually, a photometric sensor dedicated to detecting the total intensity of the light source in the front field of view is configured at the front of the vehicle, and combined with the brightness value observed by the camera, an integrated front light sequence is established. In sunny days, the common value can reach tens of thousands of lux, and at midnight without street lights, it may be less than 10 lux. First, these observed values need to be aligned in the same time slice. If the difference between multiple detection values in the same time slice does not exceed 5%, they are merged and averaged. Otherwise, additional sampling is performed in the next monitoring cycle. After the statistical process, a stable front light intensity value is obtained as , for example, if the average value of multiple detections in a certain period is about 600 lux, can be taken.

[0127] The steps for obtaining the parameters are as follows: capture the ambient light intensity through the photometric sensors installed on both sides and the roof of the vehicle. The values of these sensors can fluctuate within a certain range, but as long as the fluctuation amplitude is within 10%, they are regarded as reliable data at the same moment. Then, link and proofread these data with factors such as surrounding buildings and street lights of the vehicle to ensure that there are no extreme deviations caused by strong reflections or local shadows, and obtain the final average value of the ambient light intensity. For example, when monitoring at night on the main urban road, if the collected values fluctuate between 300 and 400 lux for multiple times, then .

[0128] The steps for obtaining the parameters are as follows: scan the obstacles through the radar or laser ranging device in front of the vehicle and record the distance data scanned each time , perform noise removal processing on the detected obstacle distances during the acquisition of the same road section, and perform a smoothing operation on the data distribution after excluding emergencies and null values. For example, moving average or taking the mean after removing extreme values can be used. If a relatively stable average distance is obtained after repeated measurements and alignment, then use it as the input value. For example, if it is found after multiple measurements that the distance between the vehicle and the vehicle in front is about 60 meters and the fluctuation range does not exceed 2 meters, then .

[0129] Calculation process:

[0130] The first step: read all the parameters. When driving on a certain section of the road, if the vehicle speed , the light influence factor , the light intensity in front , the ambient light intensity , the obstacle detection distance ;

[0131] The second step: calculate the numerator , that is ;

[0132] The third step: calculate the denominator , where , then the denominator is ;

[0133] The fourth step: calculate the first part ;

[0134] The fifth step: calculate , substitute , then ;

[0135] The sixth step: multiply the first two parts ;

[0136] Therefore, it can be obtained that ; The result shows that the comprehensive condition of the above parameters within the given time period makes the lighting mode optimization index close to 12.20. The larger the value, the more inclined the vehicle is to increase the lighting level at that moment. If the value continuously exceeds 10 in subsequent measurements, it indicates that further adjustment of the headlight brightness or angle is required. When is below 5, it indicates that the lighting demand is low, and a more energy-efficient lighting mode can be adopted.

[0137] Based on the previously obtained lighting mode optimization index, first read the vehicle steering angle, road slope, and current meteorological conditions at the same moment in chronological order. Record the steering angle value through an angle sensing device and compare it with the road curvature threshold. If the steering angle is greater than 10 degrees, it is recorded as a sharp turn situation. If the road slope exceeds 5%, it is marked as an uneven road section. Then, combined with local weather information, for example, the light attenuation coefficient may fluctuate with the increase in precipitation in rainy and snowy weather. It is necessary to compare the correlation data between the headlight illuminance and visibility under different rainy and snowy intensities in advance. When the rainfall reaches 5 millimeters per hour and the monitored light attenuation coefficient exceeds 2, it is recorded as a medium rainfall situation and the existing lighting mode is synchronously recorded accordingly. The entire process is to superimpose the steering angle, road slope, and weather attenuation information on the time series of the lighting mode optimization index, and observe whether there are significant changes between each time point one by one. If at some moments, both the steering angle and the weather attenuation coefficient increase simultaneously beyond the pre-set combined threshold, such as the angle rises above 15 degrees and the attenuation coefficient rises to 3, then the lighting level is increased by a certain amplitude for this time point and uniformly summarized into an independent execution record. If it returns to a relatively flat road section and the light attenuation is small in the subsequent period, the lighting intensity can be reduced again or the angle can be adjusted back. After summarizing the lighting parameters at each adjustment moment and marking the corresponding vehicle speed and meteorological background data, wait for the subsequent steps to apply this adjustment information to the actual headlight strategy, and finally generate the lighting mode optimization result.

[0138] The steps for obtaining the lighting performance feedback data are as follows:

[0139] Based on the lighting mode optimization result, obtain the brightness distribution of the current headlight lighting area, and combine the road reflection situation obtained by the vehicle front camera to obtain the lighting monitoring data;

[0140] According to the said lighting monitoring data, calculate the lighting performance evaluation value. The calculation formula is:

[0141]

[0142] where, represents the lighting performance evaluation value, represents the average brightness of the headlight illumination area, represents the reflection coefficient of the road surface, Represents the background brightness from the driver's perspective, Represents the monitoring time interval, Represents the average driving speed of the driver during the monitoring time, Represents the recommended driving speed in the current environment;

[0143] Based on the lighting performance evaluation value, analyze the deviation between the driver's manual adjustment behavior and the system adjustment, extract the effective lighting adjustment trend, and generate lighting performance feedback data.

[0144] Specifically, based on the lighting mode optimization result, read the brightness distribution of the current vehicle headlight illumination area. First, use the multiple sampling method to record the brightness values within the illumination range to form an initial brightness sequence, and detect whether the extremely high or low brightness data deviates from the pre-established confidence interval. For example, compare the range of 20 candela per square meter to 400 candela per square meter. If a sampling result is lower than 20, mark it as an overly dark area and check whether there is an occlusion factor. If a sampling result is higher than 400, mark it as an overly bright area and record the position of this area. Subsequently, combine these brightness data with the images captured by the front camera, and estimate the road surface reflection in front of the vehicle through the visible features on the road surface in the image. If the camera identifies multiple obvious reflective areas on the road surface, extract the coordinate ranges and light reflection intensities of these areas. After matching the time stamps of all brightness and reflection data, form corresponding data entries. If multiple groups of the same coordinates appear within a time period and the reflection intensity remains stable, determine it as a sustainable reflection area and record the corresponding brightness value. If the reflection intensity at the same position significantly decreases during other time periods, it indicates that a vehicle may have passed through this area or the lighting angle has changed. Conduct a secondary detection and comparison for this, such as taking multiple frames of data again to confirm whether it is continuous reflection or intermittent reflection. After summarizing all the brightness and reflection records collected in the above steps to form a complete time series, then compare these series with the vehicle's speed distribution and road type. If the vehicle speed shows a peak close to 70 kilometers per hour during a time period and there are more than five reflective points in front, mark this time period as a high-risk area for subsequent additional inspections. If the speed remains around 40 kilometers per hour and the number of reflective areas does not exceed two, mark it as a normal interval. After checking these records, if no data loss or abnormal fluctuations are found, consider it as an effective monitoring output and retain it within the available range. Finally, the data that can reflect the brightness distribution and road reflection information within the current vehicle headlight illumination coverage is called lighting monitoring data.

[0145] The advantage of the formula is that by introducing multiple aspects such as the average brightness of the vehicle headlight illumination, road reflection coefficient, background brightness, monitoring time, and driving speed difference, the lighting performance of the vehicle within this time interval is quantitatively measured, so as to comprehensively consider various influencing factors in the same evaluation value.

[0146] The steps for obtaining the parameter are as follows: First, set several detection points directly in front of the vehicle. Aim the photometer at the center and edge areas of the headlight irradiation area, and continuously measure the brightness values within a certain period of time to form a data set. Then, perform an extreme value removal process on the measurement results of these detection points. For example, delete the abnormal values of the smallest 1% and the largest 1%. Then sum the remaining data and divide by the number of data entries to obtain the average brightness. If the center point of the vehicle's light measures 150 candela per square meter at 10 meters, and the edge point measures 80 candela per square meter, and the measured values between other detection points mostly converge between 90 and 140 candela per square meter, then comprehensively calculate the average value of all reliable data and name it For example, finally calculate to obtain candela per square meter.

[0147] The steps for obtaining the parameter are as follows: Quantify the reflection of the road surface on the vehicle's light rays based on the detection results of the road surface reflectivity under different conditions on the same road section. Select several measurement points on the road section and use a spectral reflectometer to collect their reflectivity values respectively. If the reflectivity of some newly paved asphalt sections is commonly in the range of 0.05 to 0.25, while that of some cement road sections may be in the range of 0.20 to 0.35, classify the results of these measurement points according to the road material and take the average of the same type of road sections. If extreme values appear, they need to be re-measured and compared in the next batch of detections. Finally, obtain a unified average reflectivity coefficient representing the road surface. If the values obtained multiple times after a dozen measurements are between 0.18 and 0.22, then for example, can be selected.

[0148] The steps for obtaining the parameter are as follows: When specifically considering the background brightness from the driver's perspective, a micro-photometer needs to be installed inside the vehicle's driving cabin or near the headrest position to measure the ambient light intensity outside the driver's line of sight in front. Continuously monitor and remove interference to obtain this set of values. Then, perform a fluctuation removal on its range. If the fluctuation value exceeds 20%, secondary collection needs to be carried out in combination with the objective scene (such as suddenly entering a tunnel or the street lamp going out). After the data stabilizes, take the average of the results as For example, if the detected background brightness range mostly falls between 20 and 40 candela per square meter, then finally record it as candela per square meter.

[0149] The steps for obtaining the parameters are as follows: when setting the monitoring time, a representative duration can be selected in combination with the statistics of common vehicle driving conditions. For example, on most urban commuting roads, 120 seconds is used as a monitoring window to summarize the lighting performance. During the on-site test, the brightness and background data of the headlight illumination area are recorded once a second. After the data is collected at 120 moments, the complete evaluation process is performed. If a small number of time points are missing due to data loss or temporary failure of the measuring instrument, it can be postponed for a few seconds until 120 valid collections are made up. For example, the final result is Seconds are brought into the calculation.

[0150] The parameter acquisition step is to use the speed sensor equipment inside the vehicle to record the driver's driving speed in seconds during the monitoring time interval, and correspond it with the vehicle's driving trajectory. The speed value per second forms a sequence , and then remove the obvious abnormal points (such as instantaneous pulses on the speedometer) and take the arithmetic mean to form the average driving speed. For example, if the vehicle speed value fluctuates between 50 and 60 kilometers per hour in a certain 120-second interval, the average value can be selected Kilometers per hour .

[0151] The parameter acquisition step is to pre-set the recommended driving speed for the current environment and road conditions. It can be set at about 60 kilometers per hour on urban expressways and about 40 kilometers per hour on rural roads. This recommended speed value is set by the transportation department based on long-term observations and traffic flow distribution, and is stored locally in the vehicle information system. If the current road section is marked as an urban expressway, the recommended value can be directly read. Kilometers per hour, recorded as .

[0152] Calculation process:

[0153] The first step is to It is considered to be approximately constant during the monitoring time. To simplify:

[0154]

[0155] If you select , , , seconds, we can get:

[0156]

[0157]

[0158]

[0159] In the second step, for the denominator part perform the calculation. If the average vehicle speed is in kilometers per hour and the recommended speed is in kilometers per hour, then:

[0160]

[0161]

[0162]

[0163] In the third step, combine the numerator and the denominator:

[0164]

[0165] The result shows that the final lighting performance evaluation value is approximately -5.89. If the calculated value is greater than 0, it means that the vehicle lights relatively meet the background requirements in terms of average brightness and road reflection effect during the monitoring time. The larger the value, the more sufficient the brightness coverage and the less significant the speed difference. If the calculated value is negative, it indicates that there is a gap between the vehicle light illuminance or road reflection and the driver's background perception during this period. At the same time, the deviation of the speed from the recommended value will also pull down this evaluation value to a certain extent. The lighting level can be further optimized by combining manual adjustment or system adjustment in the subsequent links.

[0166] Based on the calculation results of the lighting performance evaluation value, first traverse the evaluation values at each time point in the entire monitoring period in sequence and compare them with the driver behavior data at the corresponding moments. If it is found that the driver manually turns down the lights at some moments while the system was originally set to high brightness, record this difference behavior. If the same situation occurs multiple times, it means that the driver has a strong personal preference for system adjustment. List these preference points in the deviation set and judge whether this manual adjustment behavior is continuous according to the frequency of its occurrence. For example, if it is found that the driver's manual adjustment actions occur 8 times within a 120 - second cycle and are all concentrated in the acceleration interval or congestion interval, then summarize this mode as the type of "tend to dim when accelerating or tend to brighten when congested". Then, combined with the corresponding lighting performance evaluation values, observe whether these adjustment actions can improve the evaluation value to be close to 0 or a more positive range. If it is found that the evaluation value changes from around -5.0 to around -2.0 after multiple manual adjustments, mark this trend as "slightly improved" in the record. If it can still maintain a fluctuation between -3.0 and -1.0 when following the subsequent period data, it is regarded as "stable adjustment". Finally, summarize all the stable deviation rules, remove individual randomly occurring adjustments, and then split and organize them one by one to obtain a set of effective lighting adjustment trends that can be used in the subsequent process. Output these final difference rules and package them as lighting performance feedback data.

[0167] Please refer toFigures 2 - 5 , the present invention also provides a vehicle lamp, including a heat dissipation lamp housing 1 with a ventilation opening 10 provided on its outer wall. A heat dissipation fan 11 corresponding to the ventilation opening 10 is installed inside the heat dissipation lamp housing 1. One end of the heat dissipation lamp housing 1 is clamped with a conductive component 2 for power supply. A copper substrate 3 is provided inside the heat dissipation lamp housing 1. LED light source chips 30 are provided on both sides of the copper substrate 3. A connection terminal 301 for electrically connecting the two LED light source chips 30 is provided at one end of the copper substrate 3. A bifurcated portion adapted to the heat dissipation fan 11 is provided at the other end of the copper substrate 3. An electronic control chip 31 is provided on one side of the copper substrate 3. Heat conduction pipe fittings 4 are provided on both outer sides of the copper substrate 3. The heat dissipation fan 11 is located between the electronic control chip 31 and the conductive component 2;

[0168] A cavity for accommodating the heat dissipation fan 11 and the conductive component 2 is formed inside the heat dissipation lamp housing 1. An opening 12 corresponding to the light-emitting surface of the LED light source chip 30 is formed on the outside of the heat dissipation lamp housing 1. The opening 12 is provided on the outside close to the ventilation opening 10.

[0169] It should be added that the conductive component 2 includes a conductive plug 21 and a rectifying plate 22. Slots a221 for welding the pins of the conductive plug 21 and slots b222 for welding the bifurcated portion of the copper substrate 3 are respectively provided around the rectifying plate 22.

[0170] It should also be added that a heat dissipation round hole 13 for the end portion of the heat conduction pipe fitting 4 to extend out is formed at the other end of the heat dissipation lamp housing 1. The other end portions of the two heat conduction pipe fittings 4 are aligned with the bifurcated portion of the copper substrate 3.

[0171] For further supplementary description, through holes corresponding to the rectifying plate 22 can be provided on the outside of the heat dissipation lamp housing, so that the opening 12 and the through holes are respectively connected to both ends of the ventilation opening 10 to form a heat dissipation air duct for hot air to flow through. When the heat dissipation fan 11 works, hot air can be discharged through the heat dissipation air duct.

[0172] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent control system, characterized in that: The following steps are involved: The environment perception module collects images of the surrounding environment through the vehicle-mounted camera, optimizes the resolution of the images, and generates optimized images; Based on the optimized image, using deep learning analysis to identify target environmental signs, including tunnels and school areas, and generating environmental sign recognition results; A light adjustment module, which adjusts the brightness of the headlights according to the environmental sign recognition result, calculates the required light parameters, and generates preliminary light adjustment parameters; based on the preliminary light adjustment parameters, further optimizes the light projection angle according to the road conditions, and generates final light adjustment parameters; A lighting mode selection module, which adjusts the final parameters based on the light, combines the current vehicle speed and weather conditions, fine-tunes the lighting mode, and generates a lighting mode optimization result; The feedback optimization module uses the lighting mode optimization result to monitor the lighting effect and driver feedback in real time, collect lighting performance data, and generate lighting performance feedback data.

2. The intelligent control system according to claim 1, characterized in that: The steps of obtaining the optimized image are: The on-board camera is started to collect a continuous image sequence of the surrounding environment, and color balance and brightness adjustment are performed to obtain a processed environment image; According to the processed environment image, a high-pass filter is applied to remove noise in the image, and edge detection is used to enhance the edge of the object in the image to obtain an environment image with optimized resolution; Based on the environment image with optimized resolution, features in the image are refined by image sharpening to generate an optimized image.

3. The intelligent control system according to claim 1, characterized in that: The steps for obtaining the environmental label recognition result are: Based on the optimized image, image features are extracted, and roads, buildings and traffic signs are separated from the background area by image segmentation to obtain candidate environmental sign areas; According to the environmental sign candidate area, a deep learning model is applied to perform classification, a feature vector is input for target matching, a matching degree is calculated and the environmental signs are screened, and an environmental sign classification result is generated; Based on the environmental sign classification results, a confidence assessment is performed to screen out misidentified targets, determine the location and category of the target environmental sign, and generate an environmental sign recognition result.

4. The intelligent control system according to claim 1, characterized in that: The steps for obtaining the preliminary light adjustment parameters are: Based on the environmental sign recognition result, determining the relationship between the recognized environmental sign type and the current position of the vehicle, calculating the real-time distance between the vehicle and the environmental sign, and obtaining light adjustment requirement data; According to the light adjustment requirement data, the required headlight brightness adjustment number is calculated, and the calculation formula is: in, Represents the required headlight brightness adjustment number, Represents the maximum setting value of the headlight brightness. Represents the minimum setting value of the headlight brightness. Represents the real-time measured distance between the vehicle and the environmental landmark, Represents the safety reference distance, Represents the height difference between the road where the vehicle is currently traveling and the environmental sign. Represents the initial angle of the current headlight projection; Based on the required headlight brightness adjustment number, real-time headlight brightness synchronization adjustment is performed to obtain preliminary light adjustment parameters.

5. The intelligent control system according to claim 1, characterized in that: The steps for obtaining the final parameters of the light adjustment are: Based on the preliminary light adjustment parameters, the light reflectivity of the road surface is detected, the reflection characteristics of the light are analyzed, the initial scattering range of the light projection is calculated, and the light projection adjustment data is obtained; According to the ray projection adjustment data, the optimized ray projection angle is calculated, and the calculation formula is: in, represents the optimized ray projection angle, Represents the height of the headlight relative to the road. Represents the height of the target area, Represents the horizontal distance between the vehicle and the target area, Represents the light reflectivity of the road surface, represents standard light reflectance; Based on the optimized light projection angle, the angle of the headlight beam is adjusted to generate a final light adjustment parameter.

6. The intelligent control system according to claim 1, characterized in that: The steps for obtaining the lighting mode optimization result are: Based on the final parameters of the light adjustment, the real-time speed information of the current vehicle is obtained, the road visibility and the surrounding environment illumination are detected, and the lighting mode adjustment data is obtained; According to the lighting mode adjustment data, the lighting mode optimization index is calculated, and the calculation formula is: in, Represents the lighting mode optimization index, Represents the current speed of the vehicle. represents the light impact factor, Represents the light intensity detected in front of the current environment. Represents the ambient light intensity around the vehicle, Represents the detection distance between the vehicle and the obstacle in front; Based on the lighting mode optimization index, a vehicle light adjustment strategy is selected to obtain a lighting mode optimization result.

7. The intelligent control system according to claim 1, characterized in that: The steps for obtaining the lighting performance feedback data are as follows: Based on the lighting mode optimization result, the brightness distribution of the current lighting area of ​​the vehicle lamp is obtained, and the lighting monitoring data is obtained by combining the road reflection situation obtained by the camera in front of the vehicle; According to the lighting monitoring data, the lighting performance evaluation value is calculated, and the calculation formula is: in, Represents the lighting performance evaluation value, Represents the average brightness of the area illuminated by the headlights. represents the reflection coefficient of the road surface, Represents the background brightness from the driver's perspective, Represents the monitoring time interval, Represents the average driving speed of the driver during the monitoring time. Represents the recommended driving speed in the current environment; Based on the lighting performance evaluation value, the deviation between the driver's manual adjustment behavior and the system adjustment is analyzed, the effective lighting adjustment trend is extracted, and the lighting performance feedback data is generated.

8. A vehicle lamp, suitable for the intelligent control system as claimed in any one of claims 1 to 7, characterized in that: include: A heat dissipation lamp housing (1) having an outer wall provided with a vent (10), a heat dissipation fan (11) corresponding to the vent (10) being arranged inside the heat dissipation lamp housing (1), a conductive component (2) for power supply being clamped at one end of the heat dissipation lamp housing (1), a copper base plate (3) being arranged inside the heat dissipation lamp housing (1), LED light source chips (30) being arranged on both sides of the copper base plate (3), a connection terminal (301) for electrically connecting two LED light source chips (30) being arranged at one end of the copper base plate (3), a bifurcation portion adapted to the heat dissipation fan (11) being arranged at the other end of the copper base plate (3), an electric control chip (31) being arranged on one side of the copper base plate (3), and heat conduction pipe fittings (4) being arranged on both sides of the outside of the copper base plate (3); The heat dissipation lamp housing (1) is provided with a cavity for accommodating the heat dissipation fan (11) and the conductive component (2), and the heat dissipation lamp housing (1) is provided with an opening (12) corresponding to the light emitting surface of the LED light source chip (30) on the outside, wherein the opening (12) is arranged outside the ventilation opening (10).

9. The vehicle lamp according to claim 8, characterized in that: The conductive component (2) comprises a conductive plug (21) and a rectifying plate (22), and the rectifying plate (22) is provided with a slot a (221) for welding the pin of the conductive plug (21) and a slot b (222) for welding the bifurcation of the copper base plate (3) on its four sides.

10. The vehicle lamp according to claim 8, characterized in that: The other end of the heat dissipation lamp housing (1) is provided with a heat dissipation circular hole (13) for one end of the heat-conducting pipe (4) to extend out, and the other ends of the two heat-conducting pipes (4) are aligned with the bifurcation of the copper base plate (3).

Images