A real-time positioning and omnibearing coverage advertisement delivery system and method

Abstract

This invention discloses a real-time positioning-based omnidirectional advertising delivery system and method, relating to the field of advertising delivery technology. The system includes a sensing unit, a trigger determination unit, an advertising processing unit, a delivery scheduling unit, and a display terminal. The sensing unit measures the location, speed, direction, road conditions, and weather of the mobile vehicle. The trigger determination unit projects the shortest interval from the location to the business district along the direction, corrects it with confidence level, and compares it with a dynamic threshold, outputting a business district number. The dynamic threshold varies with speed, road conditions, preferences, and popularity. The advertising processing unit extracts advertisements based on the business district number and ranks them by comparing four-dimensional feature vectors pairwise. The delivery scheduling unit determines the equivalent speed based on congestion status and calculates the number of advertisements allowed to be broadcast, selecting a weather-appropriate material version. The display terminal plays the advertisements, achieving adaptive and precise delivery, closed-loop effect optimization, and multi-vehicle collaborative deduplication, improving the accuracy and timeliness of advertising delivery.

Description

Technical Field

[0001] This invention relates to the field of advertising delivery technology, and in particular to a real-time, all-around coverage advertising delivery system and method. Background Technology

[0002] Mobile advertising technology is increasingly being used on taxi in-vehicle screens. Existing technologies include solutions that install positioning modules in taxis to obtain the vehicle's real-time location information and trigger advertisements from relevant business districts or merchants when the vehicle enters a pre-defined geofenced area. Other solutions employ a grid-based approach, dividing the urban area into several grids and pre-configuring a corresponding set of advertisements for each grid, then querying and playing relevant advertisements based on the taxi's current grid location. These solutions, to a certain extent, achieve location-based targeted advertising.

[0003] However, existing geofencing solutions typically set a fixed radius area centered on a business district or merchant, triggering advertising when a taxi enters this area. But in actual operation, taxi driving patterns are not constant—when vehicles are traveling at high speeds, the effective time from entering to leaving the geofence is extremely short, and triggering too late under a fixed radius will result in incomplete advertising playback; when vehicles are traveling at low speeds in congested areas, triggering too early under a fixed radius will cause the advertising content to finish playing before the passenger reaches the business district, rendering the advertisement ineffective by the time the vehicle actually enters the business district. Furthermore, existing solutions do not consider the vehicle's direction of travel. When the vehicle's direction of travel is away from the business district, even if the geographical distance is relatively short, passengers will not actually reach the business district in a short time, and triggering advertising in this situation will not effectively reach the target audience.

[0004] Furthermore, existing solutions typically employ pre-set fixed rules (such as merchant bidding levels, preset priorities, etc.) or simple carousel methods for ad ranking, failing to incorporate actual ad performance into the ranking criteria. The completion rate of the same ad may vary significantly across different time periods, road sections, and weather conditions, but existing solutions cannot dynamically adjust ad ranking weights based on these differences, leading to a rigid and unoptimized ad delivery strategy. Simultaneously, existing solutions do not consider individual passenger preferences, displaying the same ad content to passengers who prefer dining as to those who prefer shopping, resulting in insufficient personalization. On the other hand, when multiple taxis operate simultaneously in the same business district, each vehicle independently selects and decides which ads to play, potentially leading to the same ad being repeatedly played by multiple vehicles within the same timeframe, wasting advertising resources and causing audience fatigue. Summary of the Invention

[0005] To address the above issues, this invention constructs a dynamic triggering judgment and closed-loop feedback system that integrates multiple factors to achieve precise delivery and multi-vehicle collaborative optimization.

[0006] The technical solution is a real-time, location-based, all-around advertising delivery system, including:

[0007] The sensing unit measures the current geographical location, speed and direction of travel of the mobile vehicle in real time, and collects the traffic congestion index and current weather conditions of the road segment where the mobile vehicle is currently located.

[0008] The trigger determination unit calculates the shortest interval from the mobile vehicle to the boundary of each business district based on the current geographical location and the geofence information of each business district. The shortest interval is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by the confidence correction coefficient to obtain the corrected distance. The corrected distance is compared with the dynamic threshold. The dynamic threshold is the basic trigger distance plus the rate compensation term, minus the congestion attenuation term, plus the passenger preference gain term and the business district popularity gain term. When the corrected distance is less than or equal to the dynamic threshold, the corresponding business district number is output. At the same time, the predicted arrival distance and the traffic congestion index are transmitted to the deployment and scheduling unit.

[0009] The advertising processing unit extracts the corresponding set of advertisements based on the business district number. It constructs a feature vector based on four dimensions: the bid ranking score of each advertisement, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant star rating, and the degree of matching with passenger preferences. It compares the feature vectors of each advertisement in pairs and accumulates the number of wins and draws of each advertisement. The weighted sum of the number of wins and draws is used as the comprehensive advantage score. The unit outputs a list of candidate advertisements from high to low according to the comprehensive advantage score.

[0010] The delivery scheduling unit receives the predicted arrival distance and traffic congestion index, determines the equivalent delivery rate based on the movement speed and traffic congestion index, calculates the number of advertisements allowed to be broadcast in a single delivery window based on the equivalent delivery rate and predicted arrival distance, extracts the corresponding number of advertisements from the candidate advertisement list as advertisements to be broadcast, and selects the appropriate version from multiple candidate materials provided for each advertisement based on the current weather conditions.

[0011] The display terminal, installed on a mobile device, plays an adapted version of the advertisement to be broadcast.

[0012] Furthermore, the trigger determination unit calculates the trigger strength factor according to the following formula:

[0013] ;

[0014] Among them, subscript Indicates the current time, subscript Indicates the business district number. It is a natural constant;

[0015] For the mobile carrier to the business district The shortest interval of the boundary, and

[0016] ;

[0017] The current geographical location coordinates, For the business district The first geofencing The coordinates of the boundary points The direction angle is the direction of travel. The directional angle from the current position of the mobile carrier to the geometric center of the business district i. The moving speed, The traffic congestion index is the index mentioned above. The current passenger's preference density factor is determined by weighting the number and weight of the labels in the preference list. For the business district Historical advertising conversion popularity index The base trigger distance, For rate compensation coefficient, This is the congestion attenuation coefficient. This is the preference gain coefficient. This is the thermal gain coefficient. For distance confidence factor, The distance attenuation constant;

[0018] when When the mobile carrier enters the triggering area of ​​the business district, the triggering determination unit determines the triggering area of ​​the business district and outputs the business district number.

[0019] Further, the placement and scheduling unit determines the equivalent placement rate as follows: its value is the product of the movement speed and the traffic congestion index plus 1; the placement and scheduling unit calculates the number of advertisements allowed to be broadcast according to the following formula:

[0020] ;

[0021] Among them, subscript For the business district Merchant ID within;

[0022] The number of advertisements allowed to be broadcast. For the predicted arrival distance, The preset minimum safe delivery distance, The standard playback duration for a single advertisement. For merchants in the business district The completion rate of the advertisement within the previous time window. This represents the average completion rate of all advertisements within the current business district during the previous time window. This is the feedback correction factor;

[0023] The minimum safe delivery distance As the moving speed increases adaptively, its value is the base minimum safe distance plus the speed compensation term;

[0024] The ad placement and arrangement unit selects the appropriate version from multiple alternative materials provided for each advertisement based on the current weather conditions. Specifically, the material version with the highest brightness is selected when the weather is sunny, the material version with the highest color saturation is selected when the weather is rainy or snowy, and the material version with the highest contrast is selected when the weather is foggy.

[0025] Furthermore, the advertising processing unit extracts a corresponding number of advertisements from the candidate advertisement list based on the number of advertisements allowed to be broadcast output by the delivery and scheduling unit, and constructs a feature vector for each extracted advertisement. The feature vector consists of four dimensions: bidding ranking score, real-time relevance score, merchant star rating, and matching degree.

[0026] The degree of matching is determined by the cosine of the angle between the current passenger's preference vector and the targeting vector of each advertisement;

[0027] The real-time relevance score is determined by weighting and summing the actual distance between the merchant and the current location of the mobile carrier (after exponential decay) with the path planning matching degree, where the decay rate constant is... ;

[0028] The overall advantage score is obtained by comparing each of the extracted advertisements pairwise: each advertisement is compared in terms of its strengths and weaknesses in four dimensions, and the number of wins and draws of each advertisement relative to other advertisements is counted. The weighted sum of the number of wins and draws is used as the overall advantage score.

[0029] The advertising processing unit reorders the extracted advertisements from high to low according to the comprehensive advantage score.

[0030] Furthermore, it also includes a cloud collaboration device, which is connected to the delivery and scheduling unit and the display terminal on multiple mobile carriers. The cloud collaboration device collects the number of times each advertisement in the current business district has been broadcast on each mobile carrier and sends the number of broadcasts to the advertisement processing unit.

[0031] The advertising processing unit performs frequency decay correction on the overall advantage score based on the number of times it has been broadcast. The more times it has been broadcast, the greater the reduction in the overall advantage score. The corrected overall advantage score is transmitted to the placement and scheduling unit, which then redetermines the broadcast order of the advertisements to be broadcast based on the corrected overall advantage score from high to low.

[0032] Furthermore, it also includes a performance feedback unit connected to the ad placement and scheduling unit and the display terminal. The performance feedback unit records the broadcast time, broadcast position, ad number, and playback completion status of each ad broadcast and generates a log. Based on the log, it calculates the playback completion rate of each merchant's ad. Based on the deviation between the playback completion rate of each ad and the current average playback completion rate of the business district, it adjusts the real-time bidding ranking score. When the playback completion rate is higher than the average level, the corresponding real-time bidding ranking score is increased; when it is lower than the average level, the corresponding real-time bidding ranking score is decreased.

[0033] The effect feedback unit sends the corrected real-time bidding ranking score back to the advertising processing unit, and the advertising processing unit updates the corresponding dimension in the feature vector based on the corrected real-time bidding ranking score.

[0034] Furthermore, it also includes a parameter interlock verification unit, which is connected to the sensing unit and the delivery arrangement unit. The parameter interlock verification unit receives the real-time bidding ranking score corrected by the effect feedback unit, and corrects the dynamic trigger distance threshold according to the average fluctuation of the real-time bidding ranking score. The correction amount is the bidding influence coefficient multiplied by the average fluctuation.

[0035] The parameter interlock verification unit verifies the first interlock constraint relationship between the dynamic trigger distance threshold and the minimum safe delivery distance, and verifies the second interlock constraint relationship between each compensation coefficient, the basic trigger distance, the basic minimum safe distance, the maximum design rate and the bidding influence coefficient.

[0036] When the first interlock constraint is not satisfied, the parameter interlock verification unit sends a pause broadcast instruction to the broadcasting and scheduling unit; when the second interlock constraint is not satisfied, the parameter interlock verification unit sends a speed limit signal to the sensing unit.

[0037] Furthermore, it also includes a preference self-learning subunit, which is connected to the effect feedback unit and the advertising processing unit. The preference self-learning subunit obtains the playback completion rate statistically obtained by the effect feedback unit, and determines the start time of preference correction based on the verification result of the parameter interlock verification unit. Correction is initiated only when both the first interlock constraint relationship and the second interlock constraint relationship are satisfied.

[0038] The preference self-learning subunit corrects the weights of each component in the preference vector based on the deviation between the completion rate of each advertisement and the average completion rate of the current business district. Positive deviations strengthen the weights of the corresponding tag components, while negative deviations weaken the weights of the corresponding tag components. The preference self-learning subunit sends the corrected preference vector back to the advertisement processing unit, and the advertisement processing unit updates the calculation result of the matching degree based on the corrected preference vector.

[0039] A real-time, location-based, comprehensive advertising delivery method includes the following steps:

[0040] Step 1: Collect the current geographical location, speed, direction of travel, traffic congestion index, and current weather conditions of the mobile vehicle through the sensing unit;

[0041] Step 2: The trigger determination unit calculates the shortest distance from the mobile vehicle to the boundary of each business district based on the current geographical location and the geofence information of each business district. The shortest distance is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by the confidence correction coefficient to obtain the corrected distance. The corrected distance is compared with a dynamic threshold. When the corrected distance is less than or equal to the dynamic threshold, it is determined that the mobile vehicle has entered the trigger area of ​​the corresponding business district and the business district number is output.

[0042] Step 3: The advertising processing unit extracts the corresponding set of advertisements based on the business district number. It constructs a feature vector for each advertisement based on four dimensions: the bid ranking score, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant star rating, and the degree of matching with passenger preferences. The feature vectors of each advertisement are compared in pairs, and the number of wins and draws of each advertisement are accumulated. The weighted sum of the number of wins and draws is used as the comprehensive advantage score. The candidate advertisement list is output from high to low according to the comprehensive advantage score.

[0043] Step 4: The delivery scheduling unit determines the equivalent delivery rate based on the movement speed and the traffic congestion index, calculates the number of advertisements allowed to be broadcast in a single delivery window based on the equivalent delivery rate and the predicted arrival distance, extracts the corresponding number of advertisements from the candidate advertisement list as advertisements to be broadcast, and selects the appropriate version from the multiple candidate materials provided for each advertisement according to the current weather conditions.

[0044] Step 5: Play the adapted version of the advertisement to be played through the display terminal; the display terminal is a soft transparent screen installed on the inside of the rear window of the vehicle, and the content played is the advertisement and dynamic video. The displayed content is only facing out of the vehicle through the window glass on one side, and the view inside the vehicle remains transparent, so as not to affect the driver's vision.

[0045] Furthermore, in step three, the matching degree of each advertisement is determined based on the cosine of the angle between the preference vector and the targeting vector of each advertisement. The preference vector is adjusted based on the playback completion rate after each delivery. When the playback completion rate is higher than the average level of the current business district, the weight of the corresponding tag component in the preference vector is strengthened, and when it is lower than the average level, the weight of the corresponding tag component is weakened.

[0046] Due to the adoption of the above technical solutions, the present invention has the following advantages compared with the prior art;

[0047] 1. The sensing unit measures the vehicle's geographical location, speed, direction of travel, and traffic congestion index in real time. The triggering unit projects the shortest distance from the vehicle to the business district boundary along the direction of travel to obtain the predicted arrival distance, eliminating spatial deviations introduced when the direction of travel is inconsistent with the direction of the business district and avoiding false triggers when the vehicle is away from the business district. A confidence correction is applied to the predicted arrival distance. When the distance to the business district is far, the correction coefficient lowers the confidence of the predicted distance to avoid false triggers caused by prediction errors at long distances. The placement and scheduling unit also sets the equivalent placement rate to increase with the increase of the congestion index according to the congestion status, accurately reflecting the actual changes in the effective attention time of passengers when vehicles frequently start and stop under congestion conditions, so that the number of placements is matched with the actual effective exposure window. The placement and scheduling unit also selects the appropriate material version according to the weather conditions to ensure the visibility of the advertising content under different lighting and visibility conditions.

[0048] 2. The advertising processing unit constructs feature vectors for each advertisement based on four dimensions: bid ranking score, real-time correlation between the merchant and the vehicle's current location, merchant star rating, and matching degree with passenger preferences. By comparing the feature vectors of each advertisement in pairs and accumulating the number of wins and draws, a comprehensive advantage score is determined, completely avoiding the subjectivity of manually setting dimension weights and making the ranking results more objective and robust. The matching degree is determined based on the cosine of the angle between the passenger preference vector and the advertisement targeting vector, achieving personalized advertisement selection for the current passenger. The preference self-learning sub-unit corrects the weights of each component in the passenger preference vector based on the playback completion rate deviation, forming a two-dimensional differentiated feedback with the bid correction from the effect feedback unit. The cloud-based collaborative device aggregates the number of times each advertisement has been played on multiple vehicles within the same business district, downweighting high-frequency advertisements and achieving collaborative deduplication among multiple vehicles, avoiding resource waste caused by the same advertisement being repeatedly played in the same area. Attached Figure Description

[0049] Figure 1 This is a flowchart of the modules of the present invention.

[0050] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0051] The foregoing and other technical contents, features and effects of the present invention are described in conjunction with the appendix below. Figures 1 to 2 The detailed description of the embodiments will make this clear. All structural details mentioned in the following embodiments are based on the accompanying drawings.

[0052] Based on existing technology, in Embodiment 1, this embodiment provides a real-time positioning-based all-around coverage advertising delivery system, including a sensing unit, a trigger determination unit, an advertising processing unit, a delivery arrangement unit, a display terminal, a cloud collaboration device, an effect feedback unit, a parameter interlock verification unit, and a preference self-learning subunit.

[0053] The sensing unit is installed on the mobile vehicle to determine the current geographical location, speed and direction of travel of the mobile vehicle, and to collect the traffic congestion index and current weather conditions of the road segment where the mobile vehicle is currently located. The mobile vehicle is a taxi, bus or ride-hailing vehicle.

[0054] The trigger determination unit is connected to the perception unit. Based on the current geographical location and the geofence information of each business district, it calculates the shortest interval from the mobile vehicle to the boundary of each business district. The shortest interval is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by a confidence correction coefficient to obtain the corrected distance. The confidence correction coefficient decreases as the shortest interval increases. The corrected distance is compared with a dynamic threshold. The dynamic threshold is the basic trigger distance plus a rate compensation term minus a congestion attenuation term plus a passenger preference gain term and a business district popularity gain term. When the corrected distance is less than or equal to the dynamic threshold, the corresponding business district number is output. At the same time, the predicted arrival distance and traffic congestion index are transmitted to the deployment and scheduling unit.

[0055] The trigger determination unit calculates the trigger strength factor according to the following formula:

[0056] ;

[0057] Among them, subscript Indicates the current time, subscript Indicates the business district number. It is a natural constant;

[0058] For the mobile carrier to the business district The shortest interval of the boundary, and

[0059] ;

[0060] The current geographical coordinates The first geofence for business districts The coordinates of the boundary points The direction angle is the direction of travel. The directional angle from the current location of the mobile vehicle to the geometric center of the business district. For movement speed, Traffic congestion index The current passenger's preference density factor is determined by weighting the number and weight of the labels in the preference list. The historical ad conversion popularity index for business district i is determined by normalizing the click and conversion statistics of ads in that business district over the past 30 days. Based on the trigger distance, For rate compensation coefficient, This is the congestion attenuation coefficient. This is the preference gain coefficient. This is the thermal gain coefficient. For distance confidence factor, is the distance attenuation constant.

[0061] Traditional geofencing triggering methods use a fixed distance threshold, triggering ad delivery when a mobile vehicle enters that distance. This method has drawbacks because it doesn't consider factors such as the vehicle's direction of travel, speed changes, and traffic congestion. This leads to late triggering and incomplete ad playback at high speeds, and early triggering during congested, slow-moving traffic, where the ad has finished playing before the passenger reaches the shopping area. Furthermore, it's prone to false triggering when there are large errors in long-distance direction prediction. This embodiment addresses these issues by constructing a unified ratio formula for the triggering condition: the numerator is the dynamic triggering distance threshold, and the denominator is the product of the predicted arrival distance after direction correction and the confidence correction coefficient. Ad delivery is triggered when the predicted arrival distance, after confidence correction, is less than or equal to the dynamic threshold. This ensures that the triggering timing simultaneously depends on five factors: the vehicle's speed, direction, traffic conditions, passenger preferences, and the popularity of the shopping area, overcoming the limitation of a single fixed threshold in adapting to dynamic driving conditions.

[0062] Rate compensation term in molecules The faster the movement speed, the greater the trigger distance, ensuring sufficient lead time for ad playback during high-speed travel; congestion attenuation item - The trigger distance is reduced as traffic congestion increases, preventing premature triggering that would cause the advertisement to finish playing before passengers reach the shopping area; passenger preference gain item The stronger a passenger's preference for a certain type of merchant, the greater the trigger distance, thus providing priority exposure opportunities for those merchants; this is a business district popularity boosting factor. The higher the historical conversion rate of a business district, the greater the trigger distance, thus improving the overall advertising conversion effect. In the denominator... Projecting the shortest interval along the direction of travel eliminates spatial deviations introduced when the moving vehicle's direction of travel is inconsistent with the direction of the business district, preventing premature triggering when the direction of travel deviates from the business district; confidence correction coefficient. The coefficient decreases as the shortest interval increases. When the mobile vehicle is far from the business district, the reliability of the direction prediction decreases. This correction coefficient reduces the confidence of the predicted arrival distance and suppresses the trigger strength factor, thereby avoiding false triggering at long distances. When the mobile vehicle is close to the business district, the coefficient approaches 1, and the trigger judgment returns to normal sensitivity.

[0063] The value ranges from 1.0km to 2.0km. The average radius of influence of urban commercial districts usually falls within this range. If the radius is less than 1.0km, the time for mobile vehicles to traverse the commercial district is too short. If the radius is greater than 2.0km, the triggering is too early, which leads to a decrease in the relevance between the advertising information and the commercial district. The value ranges from 0.1 to 0.5. For every 10 km / h increase in moving speed, the trigger distance increases by 0.1 km to 0.5 km to ensure sufficient lead time when traveling at high speeds. The value ranges from 0.2 to 0.8, and the trigger distance decreases by 20% to 80% when the traffic congestion index increases from 0 to 1. The value ranges from 0.1 to 0.4, and the trigger distance increases by 10% to 40% as the preference density factor increases from 0 to its maximum value. The value ranges from 0.05 to 0.3, and the trigger distance increases by 5% to 30% as the historical conversion popularity index increases from 0 to its maximum value. With values ​​ranging from 0.1 to 0.5, the confidence level for long-distance direction prediction decreases by 10% to 50%. The values ​​range from 0.01 to 0.03, and the confidence correction effect gradually decays to negligible levels within the shortest interval of 1.0 km to 3.0 km.

[0064] when The time-triggered judgment unit determines when the mobile carrier enters the trigger area of ​​the business district and outputs the business district number.

[0065] The advertising processing unit is connected to the trigger determination unit, storing merchant advertising information and targeted audience tags for each business district, obtaining the current passenger's preference tags, extracting the corresponding advertising set based on the business district number output by the trigger determination unit, constructing a feature vector using four dimensions: the bidding ranking score of each advertisement, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant's star rating, and the degree of matching with the passenger's preferences, comparing the feature vectors of each advertisement in pairs and accumulating the number of wins and draws for each advertisement, using the weighted sum of the number of wins and draws as the comprehensive advantage score, and outputting a list of candidate advertisements from high to low according to the comprehensive advantage score.

[0066] Traditional ad ranking methods typically employ a weighted summation approach, assigning fixed weights to each dimension and calculating a weighted total score for ranking. This method suffers from drawbacks: the weighting is subjective, and adjusting weights becomes difficult as the number of ads increases or dimensions expand. Furthermore, it struggles to reasonably compare dimensions with different dimensions. This embodiment addresses these issues by employing a pairwise advantage ranking method: treating each ad as a point in a four-dimensional space, it compares the advantages and disadvantages of each point in each dimension, counting the number of wins and draws for each ad in all pairwise comparisons. The weighted sum of wins and draws is used as the overall advantage score for ranking. This method completely avoids the subjectivity of manually setting dimension weights, and the ranking result is unaffected by differences in dimension dimensions. As the number of ads or dimensions increases, only the number of comparisons needs to be increased, resulting in excellent scalability.

[0067] For merchants within the business district Its eigenvectors Represented as ,in For merchants The real-time bidding ranking score, For merchants The real-time correlation score between the mobile carrier and its current location. For merchants Star rating, The degree of matching is determined by the cosine of the angle between the current passenger's preference vector and the targeting vectors of each advertisement. ;

[0068] in This is the preference vector corresponding to the preference label. For merchants in the business district The targeting vector corresponds to the targeted audience tags. The real-time relevance score is determined by the weighted sum of the actual distance between the merchant and the current location of the mobile device, after exponential decay, and the path planning matching degree.

[0069] in For merchants The actual distance between the current location of the mobile carrier and the location of the mobile carrier. For merchants The path planning matching degree between the current location of the mobile vehicle and the location of the mobile vehicle. The distance decay weight ranges from 0.5 to 0.9. The decay rate constant is 0.01 to 0.05. The cumulative method for the overall advantage score is as follows: for any two ads in the extracted ad set... and Compare the values ​​of the two in each of the four dimensions, if Not lower than in all dimensions And at least in one dimension, it is higher than Then the advertisement If a winning record is obtained, Not lower than in all dimensions And at least in one dimension, it is higher than If an ad wins, it earns a winning record; if both ads have a win and a loss, they each earn a draw record. The weighted sum of the number of wins and draws for each ad in all pairwise comparisons is used as the overall advantage score. The number of wins reflects the absolute advantage of an ad in multiple dimensions, while the number of draws reflects the relative strengths and weaknesses of the ad and some other ads in certain dimensions. By using weighted summation, we can reasonably distinguish the superiority and inferiority relationships at different levels, making the ranking results more refined.

[0070] The ad delivery scheduling unit is connected to both the trigger determination unit and the ad processing unit. It receives the predicted arrival distance and traffic congestion index from the trigger determination unit, and receives the candidate ad list output by the ad processing unit. Based on the movement speed and traffic congestion index, it determines the equivalent delivery speed, which is the product of the movement speed and the traffic congestion index plus 1. ,in This is the equivalent delivery rate. In congested traffic, vehicles frequently start and stop, giving passengers more time to focus on the in-vehicle screens; therefore, the equivalent delivery rate increases with the congestion index. The number of advertisements allowed to be broadcast in the delivery scheduling unit is calculated according to the following formula:

[0071] ;

[0072] Subscript Number the merchants within the business district. The number of advertisements allowed to be broadcast. To predict the distance to be reached, The minimum safe delivery distance is preset and increases adaptively with the moving speed; its value is... ,in The minimum safety distance is based on a value ranging from 0.1km to 0.5km. The safety distance compensation coefficient ranges from 0.05 to 0.2. The standard playback duration for a single advertisement. For merchants in the business district The completion rate of the advertisement within the previous time window. This represents the average completion rate of all advertisements within the current business district during the previous time window. This is the feedback correction factor, with a value ranging from 0.05 to 0.2. The feedback correction term in this formula (1+ This allows ads with above-average completion rates to receive more play allocations, creating performance-driven advertising. The ad scheduling unit selects a certain number of ads from the candidate ad list as ads to be played, and chooses the appropriate version from multiple candidate creatives for each ad based on the current weather conditions. Specifically, the version with the highest brightness is selected when the weather is sunny, the version with the highest color saturation is selected when the weather is rainy or snowy, and the version with the highest contrast is selected when the weather is foggy. This weather-adaptive selection mechanism ensures the visibility of ad content under different lighting and visibility conditions. Ad content includes ad visuals and dynamic videos.

[0073] The display terminal is installed on a mobile carrier and connected to the delivery and arrangement unit. It receives and plays the adapted version of the advertisement to be broadcast from the delivery and arrangement unit. The display terminal is a soft transparent screen installed on the inside of the rear window of the vehicle. The content played is the advertisement and dynamic video. The display content is only visible to the outside of the vehicle through the window glass, and the view inside the vehicle remains transparent, so as not to affect the driver's vision.

[0074] The cloud-based collaborative device connects to the ad scheduling units and display terminals on multiple mobile devices, aggregating the number of times each advertisement within the current business district has been broadcast on each mobile device and sending this information to the ad processing unit. The ad processing unit adjusts the overall advantage score based on the number of broadcasts, applying frequency decay correction. The more broadcasts, the greater the reduction in the overall advantage score. The adjusted overall advantage score is then transmitted to the ad scheduling unit, which re-determines the broadcast order of the advertisements based on the adjusted overall advantage score, from highest to lowest. This cloud-based collaborative mechanism solves the problem of duplicate ad delivery that may occur when multiple mobile devices make independent decisions.

[0075] The performance feedback unit connects to the ad placement and scheduling unit and the display terminal, recording the broadcast time, location, ad number, and completion status of each ad broadcast and generating a log. Based on the log, it calculates the completion rate of each merchant's ads. The real-time bidding ranking score is adjusted based on the deviation between each ad's completion rate and the current business district's average completion rate. When the completion rate is higher than the average, the corresponding real-time bidding ranking score is increased; when it is lower than the average, the corresponding real-time bidding ranking score is decreased. The adjustment relationship is as follows:

[0076] ;

[0077] in and These are the real-time bidding ranking scores before and after the update, respectively. This is a bid adjustment coefficient ranging from 0.05 to 0.2. This adjustment increases the bid ranking score by 0.5% to 2% for every 10% increase in play completion rate above the average, matching the degree of performance deviation. The performance feedback unit sends the adjusted real-time bid ranking score back to the ad processing unit, which updates the corresponding dimension in the feature vector based on the adjusted real-time bid ranking score.

[0078] The parameter interlock verification unit is connected to the sensing unit and the delivery orchestration unit. It receives the real-time bidding ranking score corrected by the effect feedback unit and adjusts the dynamic trigger distance threshold based on the average fluctuation of the real-time bidding ranking score. The average fluctuation is the standard deviation of the bidding ranking scores over the past 10 times, and the adjustment amount is the bidding influence coefficient. Multiply by that standard deviation, The value ranges from 0.01 to 0.05. The parameter interlock verification unit verifies the dynamic trigger distance threshold. Minimum safe delivery distance The first interlocking constraint relationship between them ,in , A preset safety margin, ranging from 0.1km to 0.5km, is used. This first interlock constraint ensures that the trigger distance threshold is always greater than the minimum safe delivery distance under any dynamic conditions, guaranteeing that the mobile carrier has sufficient distance to complete the advertisement delivery after triggering, and preventing advertisement playback interruption due to insufficient margin between the trigger threshold and the safe distance. The parameter interlock verification unit also verifies the rate compensation coefficient. Safety distance compensation coefficient Congestion attenuation coefficient Preference gain coefficient Thermal gain coefficient Basic trigger distance Basic minimum safety distance Maximum design speed and bidding impact coefficient The second interlocking constraint relationship between them ,in Preference density factor The maximum value, Historical Ad Conversion Popularity Index The maximum value. This second interlock constraint ensures that the system maintains a safety margin even when all parameters take extreme values, eliminating safety risks caused by improper parameter combinations from a design perspective. When the first interlock constraint is not satisfied, the parameter interlock verification unit sends a pause broadcast command to the broadcasting and scheduling unit; when the second interlock constraint is not satisfied, the parameter interlock verification unit sends a speed limit signal to the sensing unit.

[0079] The preference self-learning subunit connects with the effect feedback unit and the advertising processing unit, obtaining the playback completion rate statistics from the effect feedback unit. Based on the verification results of the parameter interlock verification unit, it determines the timing for preference correction, initiating correction only when both the first and second interlock constraints are satisfied. This timing control mechanism ensures that preference correction will not occur when system parameters are abnormal, avoiding erroneous updates to the passenger preference model under unsafe conditions. The preference self-learning subunit adjusts the preference vector based on the deviation between the playback completion rate of each advertisement and the current average playback completion rate of the business district. The weights of each component are adjusted, with positive biases strengthening the weights of the corresponding label components and negative biases weakening the weights of the corresponding label components. The adjustment relationship is as follows:

[0080] ;

[0081] in and The first and second lines of the preference vectors before and after the correction are respectively the first line of the preference vectors before and after the correction. The weight value of each label component. The targeted tags within the business district include the first The collection of all ads for a tag. The preference learning rate ranges from 0.05 to 0.2. This preference self-learning mechanism gradually corrects its understanding of passenger preferences by accumulating playback completion rate deviations. This, combined with the bidding update mechanism of the performance feedback unit, forms a two-dimensional differentiated feedback mechanism. The performance feedback unit affects the bidding ranking score (objective dimension), while the preference self-learning sub-unit affects the passenger preference vector (subjective dimension). Neither substitutes for the other, and together they continuously improve the accuracy of ad selection. The preference self-learning sub-unit sends the corrected preference vector back to the ad processing unit, which then updates the matching degree calculation based on the corrected preference vector.

[0082] Example 2: Based on the system described in Example 1, this example provides a real-time location-based, all-around coverage advertising delivery method, which includes the following steps.

[0083] Step 1: Collect the current geographical location, speed, direction of travel, traffic congestion index, and current weather conditions of the mobile vehicle through the sensing unit.

[0084] Step 2: The trigger determination unit calculates the shortest interval from the mobile vehicle to the boundary of each business district based on the current geographical location and the geofence information of each business district. The shortest interval is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by the confidence correction coefficient to obtain the corrected distance. The corrected distance is compared with the dynamic threshold. When the corrected distance is less than or equal to the dynamic threshold, it is determined that the mobile vehicle has entered the trigger area of ​​the corresponding business district and the business district number is output.

[0085] Step 3: The advertising processing unit extracts the corresponding set of advertisements based on the business district number. It constructs a feature vector for each advertisement based on four dimensions: the bid ranking score, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant star rating, and the degree of matching with passenger preferences. The feature vectors of each advertisement are compared in pairs, and the number of wins and draws of each advertisement are accumulated. The weighted sum of the number of wins and draws is used as the comprehensive advantage score. The candidate advertisement list is output according to the comprehensive advantage score from high to low. The degree of matching of each advertisement is determined by the cosine value of the angle between the preference vector and the targeting vector of each advertisement.

[0086] Step 4: Determine the equivalent delivery rate based on the movement speed and traffic congestion index through the delivery scheduling unit. Calculate the number of advertisements allowed to be broadcast in a single delivery window based on the equivalent delivery rate and predicted arrival distance. Select the corresponding number of advertisements from the candidate advertisement list as the advertisements to be broadcast. Select the appropriate version from the multiple candidate materials provided for each advertisement based on the current weather conditions.

[0087] Step 5: Play the adapted version of the advertisement to be played on the display terminal.

[0088] The preference vector is adjusted based on the playback completion rate after each delivery. When the playback completion rate is higher than the current business district average, the weight of the corresponding tag component in the preference vector is strengthened, and when it is lower than the average, the weight of the corresponding tag component is weakened. The adjusted preference vector is used to calculate the matching degree in the subsequent step three.

[0089] In practical use, based on existing technology, the sensing unit measures the taxi's current geographical location, speed, and direction of travel in real time, and collects the traffic congestion index and weather conditions of the current road segment. While the taxi is traveling in the city, the sensing unit continuously transmits its geographical coordinates, speed, direction of travel, and traffic congestion index to the triggering and determining unit.

[0090] The trigger determination unit calculates the shortest interval based on the taxi's current geographical location and the geofence information of each business district. This interval is projected along the direction of travel to obtain the predicted arrival distance. This predicted arrival distance is then multiplied by a confidence correction coefficient to obtain the corrected distance. When the taxi is far from the business district, the reliability of the travel direction prediction decreases, and the confidence correction coefficient decreases accordingly to reduce the risk of false triggering at long distances. The trigger determination unit compares the corrected distance with a dynamic threshold. This dynamic threshold is the base trigger distance plus a rate compensation term, minus a congestion attenuation term, plus a passenger preference gain term and a business district popularity gain term. The faster the vehicle speed, the larger the rate compensation term to ensure sufficient lead time at high speeds. The more congested the road conditions, the larger the congestion attenuation term to avoid premature triggering and wasting exposure opportunities. The stronger the passenger's preference for a certain type of merchant or the higher the historical conversion rate of the business district, the larger the corresponding gain term to provide priority exposure for high-value merchants. When the corrected distance is less than or equal to the dynamic threshold, the trigger determination unit determines that the taxi has entered the trigger area of ​​that business district and outputs the business district number. Simultaneously, the predicted arrival distance and traffic congestion index are transmitted to the deployment scheduling unit.

[0091] The advertising processing unit extracts the set of advertisements from merchants within a business district based on the business district number, and simultaneously obtains the current passenger's preference tags. The unit constructs a feature vector from four dimensions: the bid ranking score of each advertisement, the real-time correlation between the merchant and the taxi's current location, the merchant's star rating, and the degree of matching with passenger preferences. Unlike existing technologies that use weighted summation and ranking by manually setting weights for each dimension, the advertising processing unit accumulates the number of wins and draws for each advertisement by comparing the feature vectors of each advertisement in pairs. It compares the merits of two advertisements one by one across the four dimensions. If one advertisement is no less than the other in all dimensions and more than the other in at least one dimension, it is recorded as a win. If both advertisements have wins and losses, they are each recorded as a draw. The weighted sum of the number of wins and draws for each advertisement in all pairwise comparisons is calculated as the overall advantage score. A list of candidate advertisements is then output in descending order of overall advantage score. This method completely avoids the subjectivity of weight setting, and the ranking result is not affected by differences in the dimensions.

[0092] After receiving the predicted arrival distance and traffic congestion index, the ad scheduling unit determines the equivalent ad delivery rate as the product of the movement speed and the traffic congestion index plus 1. In congested conditions, frequent vehicle starts and stops give passengers more time to focus on the in-vehicle screens; therefore, the equivalent delivery rate increases with the congestion index. Based on the equivalent delivery rate and predicted arrival distance, the ad scheduling unit calculates the number of ads allowed to be played within a single delivery window. A feedback correction term is incorporated into the calculation formula; ads with a completion rate higher than the average level in the business district in the previous time window receive additional playback quotas, creating a performance-driven effect. The ad scheduling unit selects a corresponding number of ads from the candidate ad list as ads to be played, and chooses the appropriate version from multiple candidate materials for each ad based on the current weather conditions: the version with the highest brightness for sunny days, the version with the highest color saturation for rainy or snowy days, and the version with the highest contrast for foggy days, ensuring the visibility of the ad content under different lighting and visibility conditions.

[0093] The display terminal is installed on the inside of the rear window of the mobile vehicle. It is a soft transparent screen that plays an adapted version of the advertisement to be broadcast. The displayed content is only visible to the outside of the vehicle through the window glass, and the view from inside the vehicle remains transparent, so as not to affect the driver's vision.

[0094] After each advertisement airs, the performance feedback unit records the airtime, location, ad number, and completion status, generating a log. Based on the log, it calculates the completion rate of each merchant's advertisements. The performance feedback unit adjusts the real-time bidding ranking score based on the deviation between each ad's completion rate and the current business district's average completion rate. If the completion rate is higher than the average, the corresponding real-time bidding ranking score is increased; if it is lower than the average, the corresponding real-time bidding ranking score is decreased. The adjusted score is then sent back to the ad processing unit, which updates the corresponding dimension in the feature vector based on the adjusted score.

[0095] The cloud-based collaborative device collects the number of times each advertisement in the current business district has been broadcast on each taxi and sends the number of broadcasts to the advertising processing unit. The advertising processing unit adjusts the overall advantage score based on the number of broadcasts. The more times the advertisement has been broadcast, the greater the reduction in the overall advantage score, in order to solve the problem of duplicate advertising that may occur when multiple taxis make independent decisions.

[0096] The parameter interlock verification unit continuously monitors the system's safety boundaries: verifying whether the dynamic trigger distance threshold is always greater than the minimum safe delivery distance, and verifying whether the constraint relationships between each compensation coefficient, basic trigger distance, basic minimum safe distance, and maximum design rate are valid, ensuring that the system can operate safely under any dynamic conditions and extreme parameter combinations. When any interlock constraint is not met, the parameter interlock verification unit sends a pause broadcast command to the delivery scheduling unit or a speed limit signal to the sensing unit.

[0097] The preference self-learning subunit only initiates preference correction when all the above safety constraints are met, avoiding erroneous updates to the passenger preference model when system parameters are abnormal. The preference self-learning subunit corrects the weights of each component in the preference vector based on the deviation between the completion rate of each advertisement and the average completion rate of the current business district. Positive deviations strengthen the weights of the corresponding label components, while negative deviations weaken the weights of the corresponding label components. The corrected preference vector is then fed back to the advertisement processing unit, which updates the matching degree calculation results based on the corrected preference vector.

[0098] The system achieves five-dimensional joint triggering through a trigger judgment unit, encompassing direction perception, speed adaptation, road condition adaptation, preference adaptation, and heat adaptation. It also achieves objective ranking without manual weight setting through an advertising processing unit, calculates the equivalent rate of congestion perception and provides feedback through a delivery scheduling unit, enables multi-vehicle collaborative deduplication through a cloud-based collaborative device, ensures safety across the entire parameter domain through a parameter interlock verification unit, and continuously optimizes passenger preferences through a preference self-learning subunit. This system overcomes multiple shortcomings of existing technologies, such as fixed threshold triggering, subjective weight ranking, independent decision-making by individual vehicles, and lack of parameter interlocking.

[0099] The above description is a further detailed explanation in conjunction with specific embodiments, and it should not be considered that the specific implementation of the present invention is limited to this. For those skilled in the art to which this invention pertains and related fields, any extensions, operation methods, and data substitutions made based on the technical solution concept of this invention should fall within the protection scope of this invention.

Claims

1. A real-time location-based, all-around coverage advertising delivery system, characterized in that, include: The sensing unit measures the current geographical location, speed and direction of travel of the mobile vehicle in real time, and collects the traffic congestion index and current weather conditions of the road segment where the mobile vehicle is currently located. The trigger determination unit calculates the shortest interval from the mobile vehicle to the boundary of each business district based on the current geographical location and the geofence information of each business district. The shortest interval is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by the confidence correction coefficient to obtain the corrected distance. The corrected distance is compared with the dynamic threshold. The dynamic threshold is the basic trigger distance plus the rate compensation term, minus the congestion attenuation term, plus the passenger preference gain term and the business district popularity gain term. When the corrected distance is less than or equal to the dynamic threshold, the corresponding business district number is output. At the same time, the predicted arrival distance and the traffic congestion index are transmitted to the deployment and scheduling unit. The advertising processing unit extracts the corresponding set of advertisements based on the business district number. It constructs a feature vector based on four dimensions: the bid ranking score of each advertisement, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant star rating, and the degree of matching with passenger preferences. It compares the feature vectors of each advertisement in pairs and accumulates the number of wins and draws of each advertisement. The weighted sum of the number of wins and draws is used as the comprehensive advantage score. The unit outputs a list of candidate advertisements from high to low according to the comprehensive advantage score. The delivery scheduling unit receives the predicted arrival distance and traffic congestion index, determines the equivalent delivery rate based on the movement speed and traffic congestion index, calculates the number of advertisements allowed to be broadcast in a single delivery window based on the equivalent delivery rate and predicted arrival distance, extracts the corresponding number of advertisements from the candidate advertisement list as advertisements to be broadcast, and selects the appropriate version from multiple candidate materials provided for each advertisement based on the current weather conditions. The display terminal, installed on a mobile device, plays an adapted version of the advertisement to be broadcast.

2. The real-time positioning-based all-around coverage advertising delivery system according to claim 1, characterized in that, The trigger determination unit calculates the trigger intensity factor according to the following formula: ； Among them, subscript Indicates the current time, subscript Indicates the business district number. It is a natural constant; For the mobile carrier to the business district The shortest interval of the boundary, and ； in, The current geographical location coordinates, For the business district The first geofencing The coordinates of the boundary points The direction angle is the direction of travel. The directional angle from the current position of the mobile carrier to the geometric center of the business district i. The moving speed, The traffic congestion index is the index mentioned above. The current passenger's preference density factor is determined by weighting the number and weight of the labels in the preference list. For the business district Historical advertising conversion popularity index The base trigger distance, For rate compensation coefficient, This is the congestion attenuation coefficient. This is the preference gain coefficient. This is the thermal gain coefficient. For distance confidence factor, The distance attenuation constant; when When the mobile carrier enters the triggering area of ​​the business district, the triggering determination unit determines the triggering area of ​​the business district and outputs the business district number.

3. The real-time positioning-based all-around coverage advertising delivery system according to claim 2, characterized in that, The placement and scheduling unit determines the equivalent placement rate as follows: its value is the product of the movement rate and the traffic congestion index plus 1; the placement and scheduling unit calculates the number of advertisements allowed to be broadcast according to the following formula: ； Among them, subscript For the business district Merchant ID within; The number of advertisements allowed to be broadcast. For the predicted arrival distance, The preset minimum safe delivery distance, The standard playback duration for a single advertisement. For merchants in the business district The completion rate of the advertisement within the previous time window. This represents the average completion rate of all advertisements within the current business district during the previous time window. This is the feedback correction factor; The minimum safe delivery distance As the moving speed increases adaptively, its value is the base minimum safe distance plus the speed compensation term; The ad placement and arrangement unit selects the appropriate version from multiple alternative materials provided for each advertisement based on the current weather conditions. Specifically, the material version with the highest brightness is selected when the weather is sunny, the material version with the highest color saturation is selected when the weather is rainy or snowy, and the material version with the highest contrast is selected when the weather is foggy.

4. The real-time positioning-based all-around coverage advertising delivery system according to claim 3, characterized in that, The advertising processing unit extracts a corresponding number of advertisements from the candidate advertisement list based on the number of advertisements allowed to be broadcast output by the placement and scheduling unit, and constructs a feature vector for each extracted advertisement. The feature vector consists of four dimensions: bidding ranking score, real-time relevance score, merchant star rating, and matching degree. The degree of matching is determined by the cosine of the angle between the current passenger's preference vector and the targeting vector of each advertisement; The real-time relevance score is determined by weighting and summing the actual distance between the merchant and the current location of the mobile carrier (after exponential decay) with the path planning matching degree, where the decay rate constant is... ; The overall advantage score is obtained by comparing each of the extracted advertisements pairwise: each advertisement is compared in terms of its strengths and weaknesses in four dimensions, and the number of wins and draws of each advertisement relative to other advertisements is counted. The weighted sum of the number of wins and draws is used as the overall advantage score. The advertising processing unit reorders the extracted advertisements from high to low according to the comprehensive advantage score.

5. The real-time positioning-based all-around coverage advertising delivery system according to claim 4, characterized in that, It also includes a cloud collaboration device, which is connected to the placement and scheduling unit and the display terminal on multiple mobile carriers. The cloud collaboration device collects the number of times each advertisement in the current business district has been broadcast on each mobile carrier and sends the number of times it has been broadcast to the advertisement processing unit. The advertising processing unit performs frequency decay correction on the overall advantage score based on the number of times it has been broadcast. The more times it has been broadcast, the greater the reduction in the overall advantage score. The corrected overall advantage score is transmitted to the placement and scheduling unit, which then redetermines the broadcast order of the advertisements to be broadcast based on the corrected overall advantage score from high to low.

6. The real-time positioning-based all-around coverage advertising delivery system according to claim 5, characterized in that, It also includes a performance feedback unit, connected to the ad placement and scheduling unit and the display terminal. The performance feedback unit records the broadcast time, broadcast position, ad number, and playback completion status of each ad broadcast and generates a log. Based on the log, it calculates the playback completion rate of each merchant's ad. Based on the deviation between the playback completion rate of each ad and the current average playback completion rate of the business district, it adjusts the real-time bidding ranking score. When the playback completion rate is higher than the average level, the corresponding real-time bidding ranking score is increased; when it is lower than the average level, the corresponding real-time bidding ranking score is decreased. The effect feedback unit sends the corrected real-time bidding ranking score back to the advertising processing unit, and the advertising processing unit updates the corresponding dimension in the feature vector based on the corrected real-time bidding ranking score.

7. The real-time positioning-based all-around coverage advertising delivery system according to claim 6, characterized in that, It also includes a parameter interlock verification unit, which is connected to the sensing unit and the delivery arrangement unit. The parameter interlock verification unit receives the real-time bidding ranking score corrected by the effect feedback unit, and corrects the dynamic trigger distance threshold according to the average fluctuation of the real-time bidding ranking score. The correction amount is the bidding influence coefficient multiplied by the average fluctuation. The parameter interlock verification unit verifies the first interlock constraint relationship between the dynamic trigger distance threshold and the minimum safe delivery distance, and verifies the second interlock constraint relationship between each compensation coefficient, the basic trigger distance, the basic minimum safe distance, the maximum design rate and the bidding influence coefficient. When the first interlock constraint is not satisfied, the parameter interlock verification unit sends a pause broadcast instruction to the broadcasting and scheduling unit; when the second interlock constraint is not satisfied, the parameter interlock verification unit sends a speed limit signal to the sensing unit.

8. The real-time positioning-based all-around coverage advertising delivery system according to claim 7, characterized in that, It also includes a preference self-learning subunit, which is connected to the effect feedback unit and the advertising processing unit. The preference self-learning subunit obtains the playback completion rate statistically obtained by the effect feedback unit and determines the start time of preference correction based on the verification result of the parameter interlock verification unit. Correction is only started when both the first interlock constraint relationship and the second interlock constraint relationship are satisfied. The preference self-learning subunit corrects the weights of each component in the preference vector based on the deviation between the completion rate of each advertisement and the average completion rate of the current business district. Positive deviations strengthen the weights of the corresponding tag components, while negative deviations weaken the weights of the corresponding tag components. The preference self-learning subunit sends the corrected preference vector back to the advertisement processing unit, and the advertisement processing unit updates the calculation result of the matching degree based on the corrected preference vector.

9. A real-time location-based, all-around coverage advertising method, applied to the system according to any one of claims 1 to 8, characterized in that, Includes the following steps: Step 1: Collect the current geographical location, speed, direction of travel, traffic congestion index, and current weather conditions of the mobile vehicle through the sensing unit; Step 2: The trigger determination unit calculates the shortest distance from the mobile vehicle to the boundary of each business district based on the current geographical location and the geofence information of each business district. The shortest distance is projected along the direction of travel to obtain the predicted arrival distance. The predicted arrival distance is then multiplied by the confidence correction coefficient to obtain the corrected distance. The corrected distance is compared with a dynamic threshold. When the corrected distance is less than or equal to the dynamic threshold, it is determined that the mobile vehicle has entered the trigger area of ​​the corresponding business district and the business district number is output. Step 3: The advertising processing unit extracts the corresponding set of advertisements based on the business district number. It constructs a feature vector for each advertisement based on four dimensions: the bid ranking score, the real-time correlation between the merchant and the current location of the mobile carrier, the merchant star rating, and the degree of matching with passenger preferences. The feature vectors of each advertisement are compared in pairs, and the number of wins and draws of each advertisement are accumulated. The weighted sum of the number of wins and draws is used as the comprehensive advantage score. The candidate advertisement list is output from high to low according to the comprehensive advantage score. Step 4: The delivery scheduling unit determines the equivalent delivery rate based on the movement speed and the traffic congestion index, calculates the number of advertisements allowed to be broadcast in a single delivery window based on the equivalent delivery rate and the predicted arrival distance, extracts the corresponding number of advertisements from the candidate advertisement list as advertisements to be broadcast, and selects the appropriate version from the multiple candidate materials provided for each advertisement according to the current weather conditions. Step 5: Play the adapted version of the advertisement to be broadcast through the display terminal.

10. The method according to claim 9, characterized in that, In step three, the matching degree of each advertisement is determined based on the cosine of the angle between the preference vector and the targeting vector of each advertisement. The preference vector is adjusted based on the playback completion rate after each delivery. When the playback completion rate is higher than the average level of the current business district, the weight of the corresponding tag component in the preference vector is strengthened, and when it is lower than the average level, the weight of the corresponding tag component is weakened.

Images