A shared electric bicycle battery replacement demand prediction method based on trajectory data mining

Abstract

The application discloses a shared electric bicycle battery replacement demand prediction method based on trajectory data mining, which comprises the following steps: 1, based on the air grid modeling, the grid function area identification is realized by combining the crawled POI data; 2, based on the actual order data, the renewable feature data is further mined according to the functional area classification, and a traffic characteristic model is established; 3, based on the battery capacity and the unit power consumption, an electric quantity characteristic model is established; 4, based on the traffic characteristic model and the electric quantity characteristic model, a shared electric bicycle battery replacement demand prediction framework is established by using the Monte Carlo simulation method. The application makes up the mismatch of the traditional prediction using the American resident survey data, and uses data driving to mine the shared electric bicycle battery replacement demand characteristics conforming to the regional characteristics from the two angles of time and space, so that the application has an important role for the operation of the shared electric bicycle and further battery replacement station site selection and capacity planning.

Description

Technical Field

[0001] This invention belongs to the field of battery swapping technology for shared electric bicycles, and particularly relates to a method for predicting the demand for battery swapping of shared electric bicycles based on trajectory data mining. Background Technology

[0002] Shared electric bicycles have attracted widespread attention as a sustainable new mode of transportation. Similar to shared bicycles, shared electric bicycles primarily serve short- to medium-distance travelers who find fixed routes unsuitable or those needing to connect with other modes of transport. Users find nearby available shared electric bicycles via their smartphones and ride between any parking spots. This sharing service improves vehicle utilization and allows users to travel at low cost. Furthermore, because shared electric bicycles are battery-powered, riding becomes more effortless and comfortable, improving accessibility and mobility. The emergence of shared electric bicycles has alleviated traffic congestion to some extent, providing residents with a greener and more environmentally friendly option for short- to medium-distance travel. Therefore, these numerous advantages have brought a booming development opportunity to the shared electric bicycle market.

[0003] However, battery issues have always been a critical factor affecting the sustainable development of shared electric bicycles. Limited battery capacity increases users' range anxiety and inhibits their willingness to use shared electric bicycles, thereby reducing the utilization rate of operating vehicles. Therefore, to ensure user availability, operators need to monitor the State of Charge (SOC) of operating vehicles and replenish the battery in a timely manner to alleviate users' range anxiety. Currently, most simulations of electric vehicles use travel patterns provided by the US Department of Transportation's NHTSA database, but in reality, travel patterns vary across different regions and modes of transportation. The emergence of big data has made accurate predictions possible. On the other hand, considering the discrete and disorderly parking of shared electric bicycles, and their travel patterns being related to functional areas, how to address the battery swapping needs of shared electric bicycles has become a major challenge in this field. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining. On the basis of functional area identification, combined with actual trajectory data, the spatiotemporal travel characteristics of shared electric bicycles are mined, and the battery swapping demand of shared electric bicycles is predicted based on spatiotemporal characteristics by constructing traffic characteristic models and power characteristic models.

[0005] Technical solution: The present invention provides a method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining. The method involves creating a spatial grid model of the target area, extracting the actual trajectory data, battery capacity, and unit power consumption of the shared electric bicycles from each spatial grid, and establishing a single-unit model of the shared electric bicycle by performing the following steps. Based on the single-unit model of the shared electric bicycle, the method predicts the battery swapping demand of the shared electric bicycles.

[0006] Step 1: Divide the target area into different spatial grids evenly according to the preset spatial scale, crawl the POI data of each spatial grid, and identify the functional attributes of each spatial grid based on kernel density analysis and frequency density method to obtain the functional area category of the target area.

[0007] Step 2: Clean and process the actual trajectory data, and extract the spatiotemporal characteristics of user usage. Extract the demand set of shared electric bicycles, state probability transition matrix, travel distance and travel speed probability density function according to the functional area categories of the target area, and construct a traffic characteristic model.

[0008] Step 3: Based on the distribution characteristics of the initial state of charge of electric vehicles, battery capacity, and power consumption per unit mileage, construct a power characteristic model for shared electric bicycles;

[0009] Step 4: After initializing the parameters for the number of shared electric bicycles in the target area, take the traffic characteristic model and the power characteristic model as inputs, and the battery swapping demand of shared electric bicycles in the target area as outputs. Use Monte Carlo simulation to construct a single model of shared electric bicycles.

[0010] Furthermore, the specific steps of step 1 are as follows:

[0011] Step 1-1: Divide the study area. In order to select an appropriate grid division scale, add sub-region labels to the origin and destination of each order according to different grid division scales, and count the amount of effective travel data under that scale.

[0012]

[0013]

[0014] In the formula: and The first The grid numbers representing the origin and destination of each order. For the first Are these orders valid travel orders with origin and destination in different regions? In order to be in Grid number at scale for Total effective travel data volume within the study area at this scale;

[0015] Steps 1-2: Obtain POI data for the study area based on online electronic maps. After data cleaning and coordinate transformation, perform secondary classification and statistics. Estimate the kernel density of POIs according to their categories to obtain the POIs within the grid. Kernel density of POI Then, weights are assigned to various POIs based on public awareness and area. The number of POIs with assigned weights. The specific calculation formula is as follows: Among them, the weight of residential POIs is 40, commercial POIs is 20, industrial POIs is 30, public service POIs is 35, and green space and plaza POIs is 45.

[0016] Steps 1-3: Determine the functional area category of the grid based on the type with the highest POI frequency density. There are five categories: residential land, commercial land, industrial land, public service land, and green space and plaza land. The formula for calculating the POI frequency density of each grid is as follows:

[0017] FD i = n i N i ( i ∈ [ 1 , 5 ] )

[0018] In the formula: For a certain grid, the first Frequency density of POI-like features For the first in the study area Total number of POIs.

[0019] Furthermore, the specific steps of step 2 are as follows:

[0020] Step 2-1: Establish the demand (OD) set for shared electric bicycles:

[0021]

[0022]

[0023]

[0024]

[0025] In the formula: and These are the sets of starting points and ending points for shared electric bicycles. and The first The starting and ending point sets of functional areas and Indicates the longitude, latitude, and time of the starting and ending points;

[0026] Step 2-2: Establish the state probability transition matrix:

[0027] Dividing the time into two-hour intervals, the state transition probability matrix at time t is obtained by counting the number of vehicles moving to another location at each time point:

[0028]

[0029] In the formula: Indicates in Within a certain time, the user leaves the function area To the functional area The probability, and That is, by functional area The probability of departing and heading to other functional areas is 1.

[0030] Steps 2-3: Based on existing travel information, calculate the average travel distance between grids. Simultaneously, the overall average travel speed follows a normal distribution.

[0031] f v ( v ) = 1 2 π s 1 v exp [ − ( v − m 1 ) 2 2 s 1 2 ]

[0032] in, , .

[0033] Furthermore, the specific steps of step 3 are as follows:

[0034] Step 3-1: Initial Battery State of Charge for First Trip Follows a normal distribution:

[0035] f SOC ( S 0 ) = 1 2 π s 2 S 0 exp [ − ( S 0 − m 2 ) 2 2 s 2 2 ]

[0036] in, =0.8, =0.1;

[0037] Step 3-2: Calculate the remaining battery state of charge of the shared electric bicycle after reaching the destination:

[0038]

[0039] In the formula: Indicates arrival at the destination The state of charge at that time; Indicates from location The state of charge at the start; The loss coefficient represents the power loss caused by factors such as road climbing during driving. This refers to the mileage traveled. Electricity consumption per kilometer; This refers to the battery capacity.

[0040] Furthermore, the specific steps of step 4 are as follows:

[0041] Step 4-1: Introduce a certain number of shared electric bicycles, and generate the initial state of charge and initial departure point grid for each shared electric bicycle using the Monte Carlo sampling method;

[0042] Step 4-2: Extract the destination grid based on the origin grid type and the state transition probability matrix, and obtain the driving mileage based on the origin grid and the destination grid.

[0043] Step 4-3: Extract the travel speed and calculate the arrival time at the destination;

[0044] Step 4-4: Calculate the remaining battery state of charge at the time of arrival at the destination based on the driving mileage and power consumption per kilometer. When the conditions are met... If the current grid is the destination grid, the battery swapping requirement is triggered; otherwise, the simulation process is repeated, using the current destination grid as the starting grid for the next iteration.

[0045] Beneficial Effects: Compared with existing technologies, this invention has the following significant advantages: This invention provides a method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining. On the one hand, it classifies shared electric bicycle trips according to different functional zones by identifying functional zones. On the other hand, it constructs a traffic characteristic model using actual trajectory data, including the demand origin-destination (OD) set, state probability transition matrix, average travel distance, and average travel speed. Based on the battery characteristic model, it simulates the battery swapping demand of shared electric bicycles using Monte Carlo simulation. This invention has strong practicality and versatility, and is of great significance for predicting the battery swapping demand of shared electric bicycles and guiding the layout of battery swapping facilities. Attached Figure Description

[0046] Figure 1 This is a flowchart of a method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining, provided by an embodiment of the present invention.

[0047] Figure 2 This is a flowchart of the Monte Carlo simulation method provided in an embodiment of the present invention. Detailed Implementation

[0048] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0049] like Figure 1 As shown, this invention provides a method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining. The specific steps are as follows:

[0050] Step 1: Divide the selected area into different spatial grids according to a certain spatial scale to complete the empty grid modeling. Using the crawled POI data, combine kernel density analysis and frequency density method to identify the functional attributes of the grid.

[0051] Step 1-1: Divide the study area. In order to select a suitable grid division scale, add a sub-region label to the origin and destination of each order according to different grid division scales, and count the amount of effective travel data under that scale.

[0052]

[0053]

[0054] In the formula: and The first The grid numbers representing the origin and destination of each order. For the first Are these orders valid travel orders with origin and destination in different regions? In order to be in Grid number at scale for Total effective travel data within the study area at this scale.

[0055] Steps 1-2: Obtain POI data for the study area based on online electronic maps, and perform secondary classification and statistics based on data cleaning and coordinate transformation. Kernel density estimation is then performed on the POIs according to their categories to obtain the POIs within the grid. Kernel density of POI Then, weights are assigned to various POIs based on public awareness and area. The number of POIs with assigned weights. The specific calculation formula is as follows: Among them, the weight of residential POIs is 40, commercial POIs is 20, industrial POIs is 30, public service POIs is 35, and green space and plaza POIs is 45.

[0056] Steps 1-3: The functional zone category of a grid is determined based on the type with the highest POI frequency density. Ultimately, five categories are defined: residential land, commercial land, industrial land, public service land, and green space and plaza land. The formula for calculating the POI frequency density of each grid is as follows:

[0057] FD i = n i N i ( i ∈ [ 1 , 5 ] )

[0058] In the formula: For a certain grid, the first Frequency density of POI-like features For the first in the study area Total number of POIs.

[0059] Step 2: Clean and process the order data, mine the spatiotemporal characteristics of user usage from the actual operation data, extract the demand set of shared electric bicycles, state probability transition matrix, travel distance and travel speed probability density function according to the functional area category, and construct a traffic characteristic model;

[0060] Step 2-1: Establish a demand (OD) set for shared electric bicycles

[0061]

[0062]

[0063]

[0064]

[0065] In the formula: and These are the sets of starting points and ending points for shared electric bicycles. and The first The starting and ending point sets of functional areas and It indicates the longitude, latitude, and time of the starting and ending points.

[0066] Step 2-2: Establish the state probability transition matrix

[0067] By dividing the time into two-hour intervals and counting the number of vehicles moving to another location at each time point, the state transition probability matrix at time t can be obtained as follows:

[0068]

[0069] In the formula: Indicates in Within a certain time, the user leaves the function area To the functional area The probability, and That is, by functional area The probability of departing and heading to other functional areas is 1.

[0070] Steps 2-3: Based on existing travel information, calculate the average travel distance between grid cells. Meanwhile, the overall average travel speed follows a normal distribution.

[0071] f v ( v ) = 1 2 π s 1 v exp [ − ( v − m 1 ) 2 2 s 1 2 ]

[0072] in, , .

[0073] Step 3: Based on the initial state of charge, battery capacity, and power consumption per unit distance of the electric vehicle, construct a power characteristic model for the shared electric bicycle;

[0074] Step 3-1: Initial Battery State of Charge for First Trip Follows a normal distribution:

[0075] f SOC ( S 0 ) = 1 2 π s 2 S 0 exp [ − ( S 0 − m 2 ) 2 2 s 2 2 ]

[0076] in, =0.8, =0.1.

[0077] Step 3-2: Calculate the remaining battery state of charge of the shared electric bicycle after reaching the destination:

[0078]

[0079] In the formula: Indicates arrival at the destination The state of charge at that time; Indicates from location The state of charge at the start; The loss coefficient represents the power loss caused by factors such as road climbing during driving. This refers to the mileage traveled. Electricity consumption per kilometer; This refers to the battery capacity.

[0080] Step Four: As Figure 2 As shown, after initializing the parameters, the traffic characteristic model and the power characteristic model are used as inputs. Monte Carlo simulation is used to construct a single model of shared electric bicycles and complete the prediction of the battery swapping demand of shared electric bicycles in the study area.

[0081] Step 4-1: Introduce a certain number of shared electric bicycles, and generate the initial state of charge and initial departure point grid for each shared electric bicycle using the Monte Carlo sampling method;

[0082] Step 4-2: Extract the destination grid based on the origin grid type and the state probability transition matrix, and obtain the driving mileage based on the origin grid and the destination grid.

[0083] Step 4-3: Extract the travel speed and calculate the arrival time at the destination;

[0084] Step 4-4: Calculate the remaining battery state of charge at the time of arrival at the destination based on the driving mileage and power consumption per kilometer. When the conditions are met... If the current grid is the destination grid, a battery swapping requirement is triggered. Otherwise, the simulation process is repeated, using the current destination grid as the starting grid for the next iteration.

[0085] It should be noted that the above description of the embodiments is only for the purpose of helping to understand the method and core idea of ​​this application. For those skilled in the art, several improvements and modifications can be made to this application without departing from the principle of this invention, and these improvements and modifications are also within the protection scope of this invention.

Claims

1. A method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining, characterized in that, Spatial grid modeling is performed on the target area, and the actual trajectory data, battery capacity and unit power consumption of shared electric bicycles are extracted from each spatial grid. By performing the following steps, a single model of shared electric bicycle is established, and the battery swapping demand of shared electric bicycles is predicted based on the single model of shared electric bicycle. Step 1: Divide the target area into different spatial grids evenly according to the preset spatial scale, crawl the POI data of each spatial grid, and identify the functional attributes of each spatial grid based on kernel density analysis and frequency density method to obtain the functional area category of the target area. Step 2: Clean and process the actual trajectory data, and extract the spatiotemporal characteristics of user usage. Extract the demand set of shared electric bicycles, state probability transition matrix, travel distance and travel speed probability density function according to the functional area categories of the target area, and construct a traffic characteristic model. The specific steps for step 2 are as follows: Step 2-1: Establish the demand (OD) set for shared electric bicycles: ； ； ； ； In the formula: and These are the sets of starting points and ending points for shared electric bicycles. and The first The starting and ending point sets of functional areas and Indicates the longitude, latitude, and time of the starting and ending points; Step 2-2: Establish the state probability transition matrix: Dividing the time into two-hour intervals, the state transition probability matrix at time t is obtained by counting the number of vehicles moving to another location at each time point: ； In the formula: Indicates in Within a certain time, the user leaves the function area To the functional area The probability, and That is, by functional area The probability of departing and heading to other functional areas is 1. Steps 2-3: Based on existing travel information, calculate the average travel distance between grids. Simultaneously, the overall average travel speed follows a normal distribution. ； in, , ; Step 3: Based on the distribution characteristics of the initial state of charge of electric vehicles, battery capacity, and power consumption per unit mileage, construct a power characteristic model for shared electric bicycles; The specific steps for step 3 are as follows: Step 3-1: Initial battery state of charge for the first trip Follows a normal distribution: ； in, =0.8, =0.1; Step 3-2: Calculate the remaining battery state of charge of the shared electric bicycle after reaching the destination: ； In the formula: Indicates arrival at the destination The state of charge at that time; Indicates from location The state of charge at the start; The loss coefficient represents the electrical energy loss caused by road climbing during the driving process; This refers to the mileage traveled. Electricity consumption per kilometer; Battery capacity; Step 4: After initializing the parameters for the number of shared electric bicycles in the target area, take the traffic characteristic model and the power characteristic model as inputs, and the battery swapping demand of shared electric bicycles in the target area as outputs. Use Monte Carlo simulation to construct a single model of shared electric bicycles.

2. The method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining according to claim 1, characterized in that, The specific steps for step 1 are as follows: Step 1-1: Divide the study area. In order to select an appropriate grid division scale, add sub-region labels to the origin and destination of each order according to different grid division scales, and count the amount of effective travel data under that scale. ； ； In the formula: and The first The grid numbers representing the origin and destination of each order. For the first Are these orders valid travel orders with origin and destination in different regions? In order to be in Grid number at scale for Total effective travel data volume within the study area at this scale; Steps 1-2: Obtain POI data for the study area based on online electronic maps. After data cleaning and coordinate transformation, perform secondary classification and statistics. Estimate the kernel density of POIs according to their categories to obtain the POIs within the grid. POI-like kernel density Then, weights are assigned to various POIs based on public awareness and area. The number of POIs with assigned weights. The specific calculation formula is as follows: Among them, the weight of residential POIs is 40, commercial POIs is 20, industrial POIs is 30, public service POIs is 35, and green space and plaza POIs is 45. Steps 1-3: Determine the functional area category of the grid based on the type with the highest POI frequency density. There are five categories: residential land, commercial land, industrial land, public service land, and green space and plaza land. The formula for calculating the POI frequency density of each grid is as follows: ； In the formula: For a certain grid, the first Frequency density of POI-like features For the first in the study area Total number of POIs.

3. The method for predicting the battery swapping demand of shared electric bicycles based on trajectory data mining according to claim 1, characterized in that, The specific steps for step 4 are as follows: Step 4-1: Introduce a certain number of shared electric bicycles, and generate the initial state of charge and initial departure point grid for each shared electric bicycle using the Monte Carlo sampling method; Step 4-2: Extract the destination grid based on the functional area type of the origin grid and the state probability transition matrix, and obtain the driving mileage based on the origin grid and the destination grid. Step 4-3: Extract the travel speed and calculate the arrival time at the destination; Step 4-4: Calculate the remaining battery state of charge at the time of arrival at the destination based on the driving mileage and power consumption per kilometer. When the conditions are met... If the current grid is the destination grid, the battery swapping requirement is triggered; otherwise, the simulation process is repeated, using the current destination grid as the starting grid for the next iteration.

Images