A parking platform berth recommendation method for peak parking demand

Abstract

The application relates to the technical field of parking service, in particular to a parking platform parking space recommendation method for peak parking demand, which comprises the following steps: obtaining historical arrival data, real-time navigation data and gate measured data of a target parking lot; constructing a dynamic prediction arrival rate; calculating a dynamic service rate of the target parking lot according to the actual time interval of multiple vehicles continuously entering in the gate measured data; calculating a dynamic entry waiting time based on the dynamic prediction arrival rate and the dynamic service rate, generating total time consumption in combination with the driving time from the user position to the target parking lot, and recommending parking spaces based on the total time consumption. Through the technical scheme, peak congestion can be predictively evaluated, the trap of not being able to enter with a parking space can be avoided, and the user parking experience during the peak period is greatly improved.

Description

Technical Field

[0001] This application relates to the field of parking service technology, and in particular to a parking platform space recommendation method for peak parking demand. Background Technology

[0002] With the surge in the number of motor vehicles in cities, the problem of parking difficulties during peak hours has become increasingly prominent. Currently, drivers generally rely on various map applications or professional parking apps to find parking spaces, and these platforms are the mainstream tools for solving this problem.

[0003] Current parking information platforms primarily recommend parking lots based on the number of remaining parking spaces and their distance from the user. However, in practical applications, especially during peak traffic hours, this recommendation method has a serious technical flaw: available spaces cannot be accessed. For example, a parking lot may show a large number of remaining spaces on the app, but a long queue may already be forming at its entrance. Traditional recommendation methods only focus on static parking space data within the parking garage, neglecting the dynamic entry efficiency outside the garage. This results in users being navigated to the parking lot but still having to wait in long lines, which not only significantly reduces the user experience but also exacerbates traffic congestion on surrounding roads.

[0004] To more accurately predict total travel time, some existing technologies introduce queuing theory models to estimate waiting times. These models rely on two key parameters: average service rate and average arrival rate. However, when applied to peak-hour scenarios, existing queuing theories suffer from static blindness due to the static nature of the average arrival rate. The average arrival rate is typically a fixed average derived from historical data, failing to reflect the rapidly changing fluctuations in real traffic flow during peak periods. While platforms may know that a large number of users are navigating and will arrive in a short time, the static average arrival rate model is unaware of this, leading to a significant underestimation of waiting times and the recommendation of incorrect parking lots to users during traffic surges. Summary of the Invention

[0005] To address the technical problem of parking difficulties caused by the serious inaccuracy of existing technologies in predicting queuing time during peak hours, this application provides a parking platform space recommendation method for peak parking demand.

[0006] This application provides a parking space recommendation method for parking platforms targeting peak parking demand, comprising: acquiring historical arrival data of the target parking lot, real-time navigation data of the platform, and measured data of the parking lot's gate; constructing a dynamic predicted arrival rate, wherein the dynamic predicted arrival rate integrates a historical arrival benchmark calculated based on the historical arrival data, a real-time traffic flow correction calculated based on the measured data of the gate, and a navigation intent pressure calculated based on the real-time navigation data; calculating a dynamic service rate of the target parking lot based on the actual time interval between multiple vehicles entering the parking lot in the measured data of the gate; calculating a dynamic entry waiting time based on the dynamic predicted arrival rate and the dynamic service rate, generating a total time consumption by combining the driving time from the user's location to the target parking lot, and recommending parking spaces based on the total time consumption.

[0007] This application constructs dynamic predicted arrival rate and dynamic service rate, so that the queuing time assessment no longer relies on rigid historical averages or ideal values, but can respond in real time to the current measured traffic flow and the actual passage efficiency of the entrance, especially the platform's unique future navigation intent, thereby achieving effective avoidance and accurate prediction of entrance congestion during peak periods.

[0008] In one embodiment, the navigation intent pressure is obtained by: setting a future perception window and a navigation user threshold; counting the total number of navigation users whose expected arrival time falls within the future perception window in real-time navigation data; and calculating the navigation intent pressure based on the total number of navigation users, the future perception window, and the navigation user threshold using a preset nonlinear amplification function, wherein when the total number of navigation users is greater than the navigation user threshold, the nonlinear amplification function amplifies the navigation intent pressure.

[0009] In one embodiment, the navigation intent pressure satisfies the following relationship: ; in, For navigation intent pressure, The total number of navigation users, A window for future perception This is the threshold for the number of people allowed to navigate.

[0010] By introducing navigation intent pressure and using the hyperbolic tangent function to nonlinearly amplify it, the static blindness defect of arrival rate is solved. It can keenly capture the influx of concentrated traffic flow and give it a punitive weight, effectively solving the technical problem that traditional models cannot predict sudden traffic flow and thus cause inaccurate predictions.

[0011] In one embodiment, the dynamically predicted arrival rate satisfies the following relationship: ; in, To dynamically predict arrival rates, As a historical benchmark, For real-time traffic flow correction, For navigation intent pressure, and These are the preset weighting coefficients.

[0012] In one embodiment, the dynamic service rate satisfies the following relationship: ; in, For dynamic service rate, This represents the actual number of vehicles within the measured window. This is the entry timestamp of the last vehicle within the measured window. This is the timestamp of the first vehicle entering the test window. This represents the minimum average service time per vehicle.

[0013] By calculating the dynamic service rate through real-time monitoring of the actual passage interval of the barrier gate, the rigidity of the ideal service rate is solved. This allows the model to respond instantly to the decline in actual passage efficiency caused by barrier gate jams, slow driver operation, etc., and avoids the shortcomings of existing technologies that use fixed values ​​and seriously underestimate queuing time. This enables adaptive evaluation of entrance service capacity.

[0014] In one embodiment, the calculation of the dynamic entry waiting time includes: calculating traffic intensity. ,in To dynamically predict arrival rates, For dynamic service rate; when the traffic intensity is less than 1, it is determined by the relational formula. Calculate dynamic entry waiting time .

[0015] In one embodiment, when the traffic intensity is not less than 1, the target parking lot is determined to be in a state of severe congestion, and a preset penalty queuing time is assigned to it as the dynamic entry waiting time.

[0016] By setting traffic intensity saturation conditions, parking lots that have exceeded their capacity in a timely manner can be identified, and punitive waiting times can be assigned to them. This effectively diverts users away from entrances that are about to be overwhelmed, preventing them from getting stuck in endlessly growing queues.

[0017] In one embodiment, the real-time traffic flow correction is obtained by statistically analyzing the total number of vehicles actually entering the gate within the most recent preset time window from the measured data of the gate, and converting it into the average number of vehicles arriving per minute.

[0018] In one embodiment, the historical arrival benchmark is obtained by extracting all vehicle entry timestamps for the same historical time period from the historical arrival data and calculating the historical average number of vehicle arrivals per minute.

[0019] In one embodiment, the real-time navigation data includes the total number of all users using the platform's navigation service whose destination is the target parking lot, as well as the estimated arrival time for each navigation user.

[0020] The technical solution of this application has the following beneficial technical effects: This application is able to move away from relying on fixed historical averages or ideal service rates, and instead incorporate the platform’s unique real-time navigation intent data into the arrival rate model, and introduce a non-linear amplification mechanism to deal with peak-hour congestion, enabling the model to predict future traffic surges.

[0021] Furthermore, by monitoring the passage interval of the barrier gate in real time to assess the actual service capacity of the entrance, an adaptive assessment of entrance efficiency is achieved. Ultimately, a more accurate dynamic entry waiting time combined with driving time is used as the basis for recommending the total time, which can effectively help users avoid parking lots where there are spaces but no one can enter, greatly improve user satisfaction during peak hours, and achieve implicit load balancing of regional parking lots. Attached Figure Description

[0022] Figure 1 This is a flowchart of a parking platform space recommendation method for peak parking demand, according to an embodiment of this application.

[0023] Figure 2 This diagram schematically illustrates a comparison between the recommended berths during peak hours using the present invention and conventional methods. Detailed Implementation

[0024] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0025] Figure 1 This is a flowchart illustrating a parking platform space recommendation method for peak-hour parking demand, according to an embodiment of this application. Figure 1 As shown, the parking platform space recommendation method for peak parking demand includes steps S101 to S104, which are described in detail below.

[0026] S101, acquire historical arrival data of the target parking lot, real-time navigation data of the platform, and actual measured data of the parking lot's barrier gate.

[0027] In one embodiment, data acquisition is performed by a parking platform server deployed in the cloud. The platform server needs to simultaneously connect to and manage multiple data streams from historical databases, real-time navigation services, and the parking lot's local gate system.

[0028] In this optional embodiment, the server extracts the sequence of vehicle entry timestamps from the platform's historical database for the target parking lot during the same historical time period. For example, if the current time is Friday evening peak at 18:30, the system extracts entry records from the 18:00-19:00 time period of the past four Fridays. This sequence is then cleaned to remove abnormal data points generated by system maintenance or testing, resulting in historical arrival data used to calculate historical benchmarks.

[0029] In this optional embodiment, the server queries the platform's navigation service module in real time to obtain real-time navigation data. This module maintains a dynamic list recording all users currently using the app for navigation whose destination is the target parking lot. The server retrieves the total number of users currently navigating and the estimated arrival time for each user from this list.

[0030] Furthermore, the server subscribes to the target parking lot's local gate management system in real time via an API interface to obtain the most recent, for example, the sequence of actual entry timestamps of all vehicles within the past 10 minutes. This data is the most direct basis for reflecting the current traffic pressure at the entrance.

[0031] Thus, by acquiring and preprocessing multi-source data, a comprehensive and real-time data foundation can be provided for the subsequent construction of dynamic arrival rate and service rate models.

[0032] S102, constructing a dynamic predicted arrival rate, which integrates the historical arrival benchmark calculated based on historical arrival data, the real-time traffic flow correction calculated based on the measured data of the gate, and the navigation intention pressure calculated based on real-time navigation data.

[0033] In one embodiment, a historical arrival benchmark can be calculated based on historical arrival data. This benchmark serves as a stable baseline representing the average traffic flow level during that time period, and is obtained by statistically analyzing the average number of vehicles arriving per minute during similar historical time periods. For example, if statistics show that an average of 2 vehicles entered the venue per minute between 18:00 and 19:00 on the past four Fridays, then the historical arrival benchmark would be 2 vehicles / minute.

[0034] In this optional embodiment, real-time traffic flow correction can be calculated based on measured data from the barrier gate. This correction is used to capture the actual traffic flow fluctuations at the current entrance, which may deviate from the historical average. It is obtained by calculating the recent actual minute arrival rate. For example, if 22 vehicles actually entered in the past 10 minutes, the real-time traffic flow correction is 2.2 vehicles / minute. As a short-term correction factor, the real-time traffic flow correction reflects the immediate pressure caused by local traffic or non-platform users not captured by the platform.

[0035] Furthermore, navigation intent pressure can be calculated based on real-time navigation data. This component is used to assess the deterministic traffic pressure that will be brought by platform navigation users in the future. Specifically, a future perception window, for example, 10 minutes, and a navigation user threshold, for example, 5 people, can be set to determine whether to activate the pressure amplification effect.

[0036] In an optional embodiment, based on the acquired real-time navigation data, all navigation users whose expected arrival time falls within the future perception window are filtered out, and their total number is counted. Then, the navigation intent pressure satisfies the following relationship: in, The pressure is for navigation intent, measured in vehicles per minute. The total number of navigation users, A window for future perception This is the threshold for the number of people allowed to navigate.

[0037] Finally, the obtained historical arrival benchmarks, real-time traffic flow corrections, and navigation intent pressure are weighted and fused to obtain the final dynamic predicted arrival rate. The dynamic predicted arrival rate satisfies the following relationship: in, To dynamically predict arrival rates, As a historical benchmark, For real-time traffic flow correction, and The preset weighting coefficients and + <1, Exemplary, The value is 0.2. The value is 0.3.

[0038] In this way, by integrating historical, real-time, and navigation intent data, the constructed dynamic prediction arrival rate becomes predictive, effectively solving the problem of inaccurate predictions caused by the inability to predict sudden traffic flow during peak hours in traditional models.

[0039] S103, calculate the dynamic service rate of the target parking lot based on the actual time interval between multiple vehicles entering the parking lot continuously in the measured data of the barrier gate.

[0040] In one embodiment, to address the problem of a fixed and rigid service rate in traditional methods, the dynamic service rate of the target parking lot can be calculated based on the actual time interval between multiple vehicles entering the parking lot continuously, as measured by the gate. The dynamic service rate satisfies the following relationship: in, For dynamic service rate, This refers to the actual number of vehicles within the measured window, for example, 10 vehicles; This is the entry timestamp of the last vehicle within the measured window. This is the timestamp of the first vehicle entering the test window. This represents the minimum average service time per vehicle, such as 0.1 minutes, or 6 seconds, which is the minimum physical time required for the barrier to be raised and for a vehicle to pass through.

[0041] For example, assuming the entrance is unobstructed, and processing the last 10 vehicles takes 4 minutes, then... =4 minutes, therefore =10 / 4.1=2.44 vehicles / minute; For congestion at the entrance, assuming it takes 6 minutes to process the last 10 vehicles due to slow driver scanning or gate malfunction, then the following can be calculated: =10 / 6.1=1.64 vehicles / minute.

[0042] In this way, by monitoring the actual passage interval of the barrier gate in real time to calculate the dynamic service rate, an adaptive assessment of the entrance service capacity is achieved, avoiding the defect of seriously underestimating the queuing time by using a fixed service rate value.

[0043] S104 calculates the dynamic entry waiting time based on the dynamically predicted arrival rate and dynamic service rate, generates the total time based on the driving time from the user's location to the target parking lot, and recommends parking spaces based on the total time.

[0044] In one embodiment, traffic intensity can be calculated based on the obtained dynamically predicted arrival rate and dynamic service rate. The dynamic entry waiting time is calculated based on the traffic intensity value.

[0045] Specifically, if A value ≥1 means that the predicted vehicle arrival rate equals or exceeds the actual processing capacity of the entrance, causing the queue to grow indefinitely. In this case, the parking lot is classified as severely congested, and the system directly assigns a very large penalty queuing time as the dynamic entry waiting time. For example, the dynamic entry waiting time... It can be set for 30 minutes, or you can remove it directly from the top of the recommended list.

[0046] If the traffic intensity is less than 1, the dynamic entry waiting time can be calculated based on the traffic intensity. The dynamic entry waiting time satisfies the following relationship: In this optional embodiment, such as Figure 2 The diagram illustrates a comparison between the proposed invention and traditional methods for recommending parking spaces during peak hours. The left diagram shows the results of the traditional recommendation method, which relies solely on historical data and distance. The right diagram shows the results of the proposed invention, which integrates navigation intent and dynamic efficiency. As can be seen in the left diagram, the total time for parking in the A-business district underground parking garage is 6.6 minutes, better than the 12.6 minutes for the B-office building parking garage. The traditional method would incorrectly recommend parking garage A. In contrast, the proposed invention maintains the same driving time baseline but predicts that parking garage A will experience longer waiting times due to traffic influx and slow entrance, ultimately causing the total time to spike to 35 minutes. Meanwhile, the total time for parking garage B remains at 12.2 minutes. Therefore, the proposed invention recommends parking garage B, which has a better total time, thus avoiding significant time waste for users.

[0047] In this way, by using the total time spent as the final recommendation criterion, users can effectively avoid parking lots that seem to have spaces but actually have long queues, solving the pain point of having spaces but not being able to enter, and greatly improving user satisfaction during peak hours.

[0048] It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application shall be determined by the appended claims.

Claims

1. A parking platform space recommendation method for peak parking demand, characterized in that, include: Acquire historical arrival data of the target parking lot, real-time navigation data from the platform, and measured data of the parking lot's barrier gate; A dynamic predicted arrival rate is constructed, which integrates the historical arrival benchmark calculated based on the historical arrival data, the real-time traffic flow correction calculated based on the measured data of the gate, and the navigation intent pressure calculated based on the real-time navigation data. The dynamic service rate of the target parking lot is calculated based on the actual time interval between multiple vehicles entering the parking lot continuously from the measured data of the gate. The dynamic entry waiting time is calculated based on the dynamic predicted arrival rate and the dynamic service rate. The total time is generated by combining the driving time from the user's location to the target parking lot. Parking spaces are recommended based on the total time. The navigation intent pressure is obtained by setting a future perception window and a navigation number threshold. The total number of navigation users whose estimated arrival time falls within the future perception window is counted from real-time navigation data. Based on the total number of navigation users, the future perception window, and a navigation user threshold, navigation intent pressure is calculated using a preset nonlinear amplification function. When the total number of navigation users exceeds the navigation user threshold, the nonlinear amplification function amplifies the navigation intent pressure. The navigation intent pressure satisfies the following relationship: in, For navigation intent pressure, The total number of navigation users, A window for future perception The threshold for the number of people navigating; the dynamically predicted arrival rate satisfies the following relationship: in, To dynamically predict arrival rates, As a historical benchmark, For real-time traffic flow correction, For navigation intent pressure, and The preset weighting coefficients; the dynamic service rate satisfies the following relationship: in, For dynamic service rate, This represents the actual number of vehicles within the measured window. This is the entry timestamp of the last vehicle within the measured window. This is the timestamp of the first vehicle entering the test window. The average minimum service time per vehicle; the calculation of the dynamic entry waiting time includes: calculating traffic intensity. ,in To dynamically predict arrival rates, For dynamic service rate; when the traffic intensity is less than 1, it is determined by the relational formula. Calculate dynamic entry waiting time When the traffic intensity is not less than 1, the target parking lot is determined to be in a state of severe congestion, and a preset penalty queuing time is assigned to it as the dynamic entry waiting time.

2. The parking platform space recommendation method for peak parking demand according to claim 1, characterized in that, The real-time traffic flow correction is obtained by statistically analyzing the total number of vehicles that actually entered the gate within the most recent preset time window from the measured data of the gate, and converting it into the average number of vehicles arriving per minute.

3. The parking platform space recommendation method for peak parking demand according to claim 1, characterized in that, The historical arrival benchmark is obtained by extracting the timestamps of all vehicles entering the venue during the same historical time period from the historical arrival data and calculating the historical average number of vehicles arriving per minute.

4. The parking platform space recommendation method for peak parking demand according to claim 1, characterized in that, The real-time navigation data includes the total number of all users using the platform's navigation service whose destination is the target parking lot, as well as the estimated arrival time of each navigation user.

Images