Bus trip destination estimation method based on network kernel density

Abstract

The application relates to the field of bus trip destination estimation, in particular to a bus trip destination estimation method based on network kernel density. The method estimates the destination of a bus trip based on network kernel density. The bus trip destination estimation method based on network kernel density comprises three steps: first, screening bus stations within a range of 440 meters of a drop-off station, and calculating the network distance from the drop-off station to the bus stations; second, calculating the network kernel density of the bus stations according to the network distance; and third, estimating the destination of the bus trip according to the network kernel density calculation result. The application can effectively estimate the destination of the bus trip.

Description

Technical Field

[0001] This invention relates to the field of public transport destination estimation, and more specifically to a public transport destination estimation method based on network kernel density. Background Technology

[0002] Kernel density estimation (KDE) is a nonparametric method for estimating the probability density function of a dataset. The basic idea is that each data point can be considered as a small probability density function (kernel), and by superimposing these kernel functions, the probability density function of the entire dataset can be approximated. Kernel density estimation has proven to be effective in clustering large-scale spatial point datasets and can reveal travel behavior characteristics and land use patterns from the results. Kernel density estimation is divided into Euclidean distance-based search and network distance-based search. Euclidean distance-based kernel density estimation may include many public transport stops outside the actual window width search range, leading to an overestimation. In network distance-based search, the focus is on the connectivity and distances between nodes in the network, rather than traditional geometric distances.

[0003] Current methods for estimating public transport destinations include statistical methods based on historical data, machine learning methods, and methods based on mobile device data. Statistical methods based on historical data analyze and model passenger travel history. They calculate the frequency or probability of travel from different boarding points to various alighting points and predict the probability distribution of passenger alighting points based on the characteristics of the current boarding point. Machine learning methods use features such as passenger boarding point, time, and weather as input, and the destination point as output, to train a regression or classification model to predict passenger destinations. Methods based on mobile device data utilize location data from passengers' mobile devices for destination estimation. They predict passenger destinations by analyzing passenger movement trajectory data and travel behavior patterns. These methods generally require massive amounts of multi-source data to estimate public transport destinations, which undoubtedly increases computational costs. Currently, there is no simple and effective method for estimating public transport destinations. Summary of the Invention

[0004] This invention provides a method for estimating public transport travel destinations based on network kernel density. By filtering public transport stops near the drop-off point and calculating the network distance from the drop-off point to these stops, the network kernel density is calculated based on the network distance of the public transport stops. The network kernel density values ​​are summed at the traffic analysis area scale to obtain the probability that public transport passengers will travel to the traffic analysis area, thereby achieving the estimation of public transport travel destinations.

[0005] Based on the above description, the present invention provides a public transport destination estimation method based on network kernel density, comprising the following steps.

[0006] Step 1: Filter bus stops within 440 meters of the drop-off point and calculate the network distance from the drop-off point to these stops.

[0007] First, select a bus stop, and then search for all bus stops within a 440-meter radius of that bus stop. i (i = 1, 2, ..., n);

[0008] Secondly, calculate the network distance d(stop,stop) from the bus stop to the searched bus stop. i );

[0009] Repeat the above steps until all bus stops have been visited.

[0010] Step 2: Calculate the network kernel density of the above bus stops based on the network distance.

[0011] For the network distance d(stop,stop) obtained in the first step i ), calculate the network kernel density value Kstop from bus stop to bus stops within the search range. i The calculation formula is as follows:

[0012]

[0013] In the formula, h is the value that minimizes the mean square integral error function of the kernel function, and k[·] is the kernel function.

[0014] h is based on the formula

[0015]

[0016] The calculation yields the following values: 1.06 is an empirical value, c is the standard deviation of the network distance from the drop-off point to the searched point, and n is the number of people who get off at the drop-off point. In this embodiment, the scenario is 1 person, i.e., n = 1.

[0017] Step 3: Estimate the destination of public transport trips based on the network kernel density calculation results.

[0018] The network kernel density values ​​of bus stops belonging to the same traffic zone are summed to obtain the probability of a bus trip destination.

[0019] Step 4: Destination estimation process ends.

[0020] The advantages and beneficial effects of this invention are as follows:

[0021] This invention proposes a public transport destination estimation method based on network kernel density. The method first filters public transport stops within a 440-meter radius of the drop-off point and calculates the network distance from the drop-off point to these stops. Secondly, it calculates the network kernel density of these public transport stops based on the network distance, and estimates the public transport destination based on the calculation results. This method fully considers the spatial distribution characteristics of adjacent public transport stops. By estimating the kernel density of the area surrounding the drop-off point, kernel density values ​​for different public transport stops can be obtained. The probability obtained by summing the kernel density values ​​at the traffic analysis area scale reflects the tendency of public transport passengers to travel to that traffic analysis area. Attached Figure Description

[0022] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0023] Figure 1 A flowchart of a public transport destination estimation method based on network kernel density provided in this application embodiment;

[0024] Figure 2 Examples of searched bus stops and calculated network distances provided in this application embodiment;

[0025] Figure 3 The calculation results of the bus stop network kernel density provided in the embodiments of this application.

[0026] Figure 4 The public transport destination estimation results provided in the embodiments of this application.

[0027] Figure 5 A schematic diagram showing the geographical units into which the study area is divided to meet the needs of traffic analysis and traffic demand forecasting. Detailed Implementation

[0028] The technical solution of the present invention will now be described clearly and in detail with reference to the accompanying drawings. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0029] Methods for estimating public transport travel destinations based on network kernel density include:

[0030] Step 1: Filter bus stops within the drop-off area and calculate the network distance from the drop-off point to these stops; filter bus stops within a 440-meter radius of the drop-off point; first, select a bus stop, and then search for bus stops within a 440-meter radius of that bus stop. i (i = 1, 2, ..., n);

[0031] Secondly, calculate the network distance d(stop,stop) from the bus stop to the searched bus stop.i );

[0032] Repeat the above steps until all bus stops have been visited.

[0033] Step 2: Calculate the network kernel density of the aforementioned bus stops based on the network distance; for the network distance d(stop,stop) obtained in Step 1; i ), calculate the network kernel density value Kstop from bus stop to bus stops within the search range. i The calculation formula is as follows:

[0034]

[0035] In the formula, h is the value that minimizes the mean square integral error function of the kernel function, and k[·] is the kernel function;

[0036] Step 3: Estimate the destination of the public transport trip based on the calculation results of Step 2; sum the network kernel density values ​​of public transport stops belonging to the same traffic analysis area, and the resulting value is the probability of the public transport trip destination.

[0037] In this embodiment of the invention, taking the destination estimation of a bus passenger as an example, the data used includes bus stop data, traffic analysis area data, and road network data. The traffic analysis area is a geographical unit into which the study area is divided, combining traffic analysis and traffic demand forecasting needs. It is the smallest spatial unit for analyzing resident and vehicle travel and distribution, such as... Figure 5 As shown in the figure.

[0038] The specific implementation of this embodiment of the invention includes the following steps:

[0039] Step 1: Select bus stops within 440 meters of the bus passenger alighting point and calculate the network distance from the alighting point to these stops. The 440-meter distance is chosen based on literature findings and refers to the average walking distance for bus passengers after alighting.

[0040] Taking a passenger alighting at bus stop C5 as an example, firstly, ArcGIS's nearest neighbor analysis tool was used to calculate the number of bus stops within a 440-meter radius of bus stop C5, selecting the shortest path. A total of 87 bus stops were found within the 440-meter search radius of bus stop C5. Secondly, Python was used to call the OpenStreetMap API to calculate the network distances from bus stop C5 to the 87 found bus stops. The bus stops within the 440-meter search radius of bus stop C5 and their network distances to bus stop C5 are shown below. Figure 2 As shown.

[0041] Step 2: Calculate the network kernel density of the above bus stops based on the network distance.

[0042] The standard deviation of the network distance calculated in step one is 124.183 meters. The formula for calculating the optimal bandwidth is:

[0043]

[0044] In the formula, c is the standard deviation of the network distance of the bus stops found in step one, and n is the number of bus passengers. The calculated optimal bandwidth is 131.63. Based on previous experience, the linear length is taken as 1 / 10 of the optimal bandwidth, i.e., the optimal bandwidth is 13.163 meters. The kernel density values ​​for the aforementioned bus stops are calculated based on the optimal bandwidth, and the calculated network kernel density values ​​are as follows: Figure 3 As shown.

[0045] Step 3: Estimate the destination of public transport trips based on the network kernel density calculation results.

[0046] The network kernel density values ​​of bus stops belonging to the same traffic analysis zone are summed to obtain the probability that bus passengers will travel to different traffic analysis zones after disembarking. The calculation results are as follows: Figure 4 As shown. In this embodiment, the destination with the highest probability for bus passengers is the traffic analysis area to which the alighting station belongs.

[0047] The names of bus stops included in the traffic analysis area, kernel density values, and summation results (probability of traveling to the traffic analysis area) are presented in tabular form.

[0048]

[0049] Step 4: Destination estimation process ends.

[0050] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A public transport destination estimation method based on network kernel density, characterized in that, include: Step 1: Filter bus stops within 440 meters of the drop-off point and calculate the network distance from the drop-off point to these stops; Step 2: Calculate the network kernel density of the above bus stops based on the network distance; Step 3: Sum the network kernel density values ​​of bus stops belonging to the same traffic analysis area. The resulting value is the probability of the bus trip destination. The traffic analysis area is a geographical unit into which the study area is divided, combining the needs of traffic analysis and traffic demand forecasting. It is the smallest spatial unit for analyzing the travel and distribution of residents and vehicles.

2. The public transport destination estimation method based on network kernel density according to claim 1, characterized in that: The first step includes: First, selecting a bus stop. Search for bus stops within a 440-meter radius of the current bus stop. .

3. The public transport destination estimation method based on network kernel density according to claim 2, characterized in that: The first step also includes: secondly, calculating bus stops. Network distance to the searched bus stop ; Repeat the above steps until all bus stops have been visited.

4. The public transport destination estimation method based on network kernel density according to claim 1, characterized in that: The second step includes: for the network distance obtained in the first step Calculate bus stops separately Network kernel density values ​​of bus stops within the search range .

5. The public transport destination estimation method based on network kernel density according to claim 4, characterized in that: Calculate bus stops Network kernel density values ​​of bus stops within the search range The calculation formula is as follows: In the formula, To minimize the mean square integral error function of the kernel function, This is the kernel function.

Images