A short-term passenger flow prediction method for new line stations in urban rail transit

Abstract

The present invention relates to a method for predicting short-term passenger flow at a new line station of urban rail transit, comprising: 1. collecting multi-source data of a new line station, the multi-source data including land use properties, traffic supply level, land development intensity and existing urban rail transit passenger flow; 2. processing and integrating the collected multi-source data to construct a passenger flow prediction model; cleaning and preprocessing the passenger flow data; obtaining the passenger flow prediction result of the new station of the urban rail by calculating the influencing factors of the passenger flow data, the passenger flow data including station attraction and passenger own factors; constructing a BP neural network model according to the passenger flow data and POI data; obtaining the short-term passenger flow of the new line station according to the constructed BP neural network model; 3. optimizing the constructed passenger flow prediction model, correcting the passenger flow and performing passenger flow distribution. The present invention can accurately predict the short-term passenger flow of the new line, which is convenient for the operation organization of the urban rail.

Description

Technical Field

[0001] The invention relates to a method for predicting short-term passenger flow at a new line station of urban rail transit, and belongs to the technical field of urban rail transit. Background Art

[0002] With the development of social economy, urban population is increasing, urban scale and scope are expanding, and residents' travel needs and travel distances are also growing sharply. In order to meet the travel needs of residents, many large cities are using large-capacity, long-distance, and high-efficiency rail transit as the backbone system for residents' public transportation. With the extension of new lines and the opening of new stations, urban rail transit has entered the network operation stage. The connection between lines is closer, and the impact and influence of new line access on the passenger flow of the existing line network is more complex. The passenger flow forecast of new lines is particularly critical to the adjustment of the transportation organization of the existing line network.

[0003] In the actual operation process, after the opening of new lines and stations, the passenger flow will go through the cultivation period, growth period and maturity period. The growth and fluctuation patterns of passenger flow in different stages are different, resulting in poor accuracy in short-term passenger flow prediction. The operation organization is constrained by the lack of accurate short-term passenger flow data support, resulting in the inability of peak-time transport capacity to meet demand, and waste of off-peak capacity, low transportation efficiency, and difficulty in meeting the requirements of refined operation management. Therefore, short-term passenger flow prediction for new lines is an important task for the current urban rail operation organization, but the robustness and reliability of traditional passenger flow prediction methods are insufficient, and the prediction accuracy needs to be further improved. Summary of the invention

[0004] The purpose of the present invention is to overcome the problems existing in the prior art and provide a method for predicting short-term passenger flow at new line stations of urban rail transit, which can accurately predict the short-term passenger flow of new lines and facilitate the operation organization of urban rail.

[0005] A method for predicting short-term passenger flow at a new line station of urban rail transit according to the present invention comprises the following steps:

[0006] Step 1: Collect multi-source data of new line stations, including land use characteristics, transportation supply level, land development intensity and passenger flow of existing urban rail transit;

[0007] Step 2: Process and integrate the collected multi-source data to build a passenger flow prediction model;

[0008] Step 2.1: Clean and preprocess the passenger flow data;

[0009] Step 2.2: Obtain the passenger flow forecast result of the new urban rail station by calculating the influencing factors of the passenger flow data, wherein the passenger flow data includes station attractiveness and passenger factors;

[0010] Step 2.3: Construct a BP neural network model based on passenger flow data and POI data;

[0011] Step 2.4: Obtain the short-term passenger flow of the new line station based on the constructed BP neural network model;

[0012] Step 3: Optimize the constructed passenger flow prediction model, correct the short-term passenger flow of the new line station obtained in step 2.4, and distribute the passenger flow.

[0013] Furthermore, the station attractiveness described in step 2 is calculated as follows:

[0014] A1: Calculate the probability of passengers going to the new station by walking, cycling and taking the bus respectively. The formula is as follows:

[0015]

[0016] In the formula, d(S i , r m ) is station S i The area covered by the attraction range is r m The straight-line distance between The three attraction ranges are walking, cycling and bus. m area;

[0017] A2: Calculate the relative probability of choosing walking, cycling or bus as the three transportation modes for continuous transfer. The formula is as follows:

[0018]

[0019] In the formula, P′ walk (d(S i , r m ))、P′ bike (d(S i , r m ))、P′ bus (d(S i , r m )) respectively represent when the distance is d(S i , r m ) when the relative probability of continuous transfer;

[0020] A3: Calculate the attraction range coverage area r m For Station S i The membership degree σ(S i , r m ), the calculation formula is as follows:

[0021]

[0022] According to the membership degree σ(Si , r m ) to get station S i The attraction of x k , the formula is as follows:

[0023] In the formula, is a region with property k.

[0024] Furthermore, the POI data is classified as follows:

[0025] (1) Non-mandatory travel POI data: including catering services, auto repair services, sports and leisure, scenic spots, and accommodation services;

[0026] (2) Mandatory travel POI data: including companies, scientific research institutes, government organizations, financial services, and various schools;

[0027] (3) Residential POI data: commercial residences.

[0028] Furthermore, step 2.4 specifically includes the following steps:

[0029] S2.41: Establish BP neural network structure;

[0030] S2.42: Establish an improved particle swarm algorithm and determine the size of the particle population, the number of particle iterations, the learning factor parameters, and the limit range of the inertia weight parameters;

[0031] S2.43: The particle fitness is obtained by using the difference between the predicted value and the actual value of the neural network, and the current fitness is compared with the optimal fitness of the particle before. If the current fitness is higher than the optimal fitness obtained before, the extreme value of the particle is updated, otherwise the original extreme value of the particle is maintained; after completing the fitness update, the global optimal value is updated;

[0032] S2.44: Calculate and update the particle speed and position according to the inertia weight formula and the particle speed and position update formula. If the updated particle speed and position exceed the upper limit, the upper limit is used as the update value. Otherwise, if the updated particle speed and position exceed the lower limit, the lower limit is used as the update value.

[0033] S2.45: Judge the result after iteration. If the termination condition is met, the iteration result is taken as the optimal solution and jump to step 2.46. Otherwise, the termination condition is not met and the iteration is continued.

[0034] S2.46: Complete the particle swarm iteration, use the obtained global optimal solution as the initial weight of the BP neural network, and train the BP neural network model.

[0035] 4. The method for predicting short-term passenger flow at new line stations of urban rail transit according to claim 1 is characterized in that: the BP neural network structure in S2.41 includes an input layer, an output layer and a hidden layer, the hidden layer is one or more layers, and the relationship between the number of neurons q in the hidden layer and the number of neurons d in the input layer is: q = 2d + 1; the data in the input layer is the attraction x of the station k , and sum the passenger flow of the station according to the POI classification data, and the data of the output layer is the passenger flow of the station.

[0036] Furthermore, the cleaning and preprocessing in step 2.1 includes eliminating data information of roads or stations that are inaccessible in the short term due to construction, and data information of roads or stations that have been abolished or are about to be abolished due to the opening of new lines.

[0037] Furthermore, the calculation formula of the learning factor parameter is as follows:

[0038]

[0039] Where: are the initial values ​​of the learning factors c1 and c2|, is the final iteration value of the learning factors c1 and c2|, t is the current iteration number, and T is the maximum iteration number;

[0040] The calculation formula of the limit range of the inertia weight parameter is as follows:

[0041] ω(τ)=ω1-(ω1-ω2)(τ / T) 2

[0042] Where ω(τ) is the inertia weight at time τ, ω1 and ω2 are the inertia weights at the initialization and end of the particle swarm algorithm, respectively.

[0043] The beneficial effects of the present invention are as follows: this paper further clarifies the changing characteristics and influencing factors of passenger flow through the analysis of passenger flow characteristics of urban rail transit under networking conditions and the research on passenger flow prediction methods for the opening of new lines, improves the accuracy of passenger flow prediction for newly built stations and existing stations in the initial stage of the opening of new lines, and has a good guiding significance for optimizing the overall management efficiency of rail transit in the initial stage of the opening of new lines and improving the quality of passenger service. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a flow chart of a method for predicting short-term passenger flow at a new line station of urban rail transit. DETAILED DESCRIPTION

[0045] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0046] like Figure 1 As shown, the short-term passenger flow prediction method for new line stations of urban rail transit of the present invention comprises the following steps:

[0047] A method for predicting short-term passenger flow at a new line station of urban rail transit according to the present invention comprises the following steps:

[0048] Step 1: Collect multi-source data of new line stations, including land use characteristics, transportation supply level, land development intensity and existing urban rail transit passenger flow;

[0049] Step 2: Process and integrate the collected multi-source data to build a passenger flow prediction model;

[0050] Step 2.1: Clean and preprocess the passenger flow data; the cleaning and preprocessing in step 2.1 includes eliminating data information of roads or stations that are inaccessible in the short term due to construction, and data information of roads or stations that have been abolished or are about to be abolished due to the opening of new lines.

[0051] Step 2.2: Obtain the passenger flow forecast result of the new urban rail station by calculating the factors affecting the passenger flow data. The passenger flow data includes the station attraction and the passenger's own factors;

[0052] The station attractiveness in step 2 is calculated as follows:

[0053] A1: Calculate the probability of passengers going to the new station by walking, cycling and taking the bus respectively. The formula is as follows:

[0054]

[0055] In the formula, d(S i , r m ) is station S i The area covered by the attraction range is r m The straight-line distance between The three attraction ranges are walking, cycling and bus. m area;

[0056] A2: Calculate the relative probability of choosing walking, cycling or bus as the three transportation modes for continuous transfer. The formula is as follows:

[0057]

[0058] In the formula, P′ walk (d(Si ,r m ))、P′ bike (d(S i , r m ))、P′ bus (d(S i , r m )) respectively represent when the distance is d(S i ,r m ) when the relative probability of continuous transfer;

[0059] A3: Calculate the attraction range coverage area r m For Station S i The membership degree σ(S i , r m ), the calculation formula is as follows:

[0060]

[0061] According to the membership degree σ(S i , r m ) to get station S i The attraction of x k , the formula is as follows:

[0062] In the formula, is a region with property k.

[0063] Step 2.3: Construct a BP neural network model based on passenger flow data and POI data;

[0064] The classification of POI data is as follows:

[0065] (1) Non-mandatory travel POI data: including catering services, auto repair services, sports and leisure, scenic spots, and accommodation services;

[0066] (2) Mandatory travel POI data: including companies, scientific research institutes, government organizations, financial services, and various schools;

[0067] (3) Residential POI data: commercial residences.

[0068] Step 2.4: Obtain the short-term passenger flow of the new line station based on the constructed BP neural network model;

[0069] S2.41: Establish a BP neural network structure; the BP neural network structure in S2.41 includes an input layer, an output layer and a hidden layer, the hidden layer is one or more layers, and the relationship between the number of neurons q in the hidden layer and the number of neurons d in the input layer is:

[0070] q=2d+1; the data of the input layer is the attraction x of the station k, and sum the passenger flow of the station according to the POI classification data, and the data in the output layer is the passenger flow of the station.

[0071] S2.42: Establish an improved particle swarm algorithm and determine the size of the particle population, the number of particle iterations, the learning factor parameters, and the limit range of the inertia weight parameters;

[0072] The calculation formula of the learning factor parameter is as follows:

[0073]

[0074] Where: are the initial values ​​of the learning factors c1 and c2|, is the final iteration value of the learning factors c1 and c2|, t is the current iteration number, and T is the maximum iteration number;

[0075] The calculation formula for the limit range of the inertia weight parameter is as follows:

[0076] ω(τ)=ω1-(ω1-ω2)(τ / T) 2

[0077] Where ω(τ) is the inertia weight at time τ, ω1 and ω2 are the inertia weights at the initialization and end of the particle swarm algorithm, respectively.

[0078] S2.43: The particle fitness is obtained by using the difference between the predicted value and the actual value of the neural network, and the current fitness is compared with the optimal fitness of the particle before. If the current fitness is higher than the optimal fitness obtained before, the extreme value of the particle is updated, otherwise the original extreme value of the particle is maintained; after completing the fitness update, the global optimal value is updated;

[0079] S2.44: Calculate and update the particle speed and position according to the inertia weight formula and the particle speed and position update formula. If the updated particle speed and position exceed the upper limit, the upper limit is used as the update value. Otherwise, if the updated particle speed and position exceed the lower limit, the lower limit is used as the update value.

[0080] S2.45: Judge the result after iteration. If the termination condition is met, the iteration result is taken as the optimal solution and jump to step 2.46. Otherwise, the termination condition is not met and the iteration is continued.

[0081] S2.46: Complete the particle swarm iteration, use the obtained global optimal solution as the initial weight of the BP neural network, and train the BP neural network model.

[0082] Step 3: Optimize the constructed passenger flow prediction model, correct the short-term passenger flow of the new line station obtained in step 2.4, and distribute the passenger flow.

[0083] The present invention can provide data support for subsequent rail line planning and provide a reference basis for urban transportation planning, public transportation system construction and rail transit network layout.

[0084] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for predicting short-term passenger flow at a new line station of urban rail transit, characterized in that: The steps include: Step 1: Collect multi-source data of new line stations, including land use characteristics, transportation supply level, land development intensity and passenger flow of existing urban rail transit; Step 2: Process and integrate the collected multi-source data to build a passenger flow prediction model; Step 2.1: Clean and preprocess the passenger flow data; Step 2.2: Obtain the passenger flow forecast result of the new urban rail station by calculating the influencing factors of the passenger flow data, wherein the passenger flow data includes station attractiveness and passenger factors; Step 2.3: Construct a BP neural network model based on passenger flow data and POI data; Step 2.4: Obtain the short-term passenger flow of the new line station based on the constructed BP neural network model; the specific steps are as follows: S2.41: Establishing a BP neural network structure; the BP neural network structure includes an input layer, an output layer and a hidden layer, the hidden layer is one or more layers, and the relationship between the number of neurons q in the hidden layer and the number of neurons d in the input layer is: q=2d+1; S2.42: Establish an improved particle swarm algorithm and determine the size of the particle population, the number of particle iterations, the learning factor parameters, and the limit range of the inertia weight parameters; S2.43: The particle fitness is obtained by using the difference between the predicted value and the actual value of the neural network, and the current fitness is compared with the optimal fitness of the particle before. If the current fitness is higher than the optimal fitness obtained before, the extreme value of the particle is updated, otherwise the original extreme value of the particle is maintained; after completing the fitness update, the global optimal value is updated; S2.44: Calculate and update the particle speed and position according to the inertia weight formula and the particle speed and position update formula. If the updated particle speed and position exceed the upper limit, the upper limit is used as the update value. Otherwise, if the updated particle speed and position exceed the lower limit, the lower limit is used as the update value. S2.45: Judge the result after iteration. If the termination condition is met, the iteration result is taken as the optimal solution and jump to step 2.

46. Otherwise, the termination condition is not met and the iteration is continued. S2.46: completing the particle swarm iteration, taking the obtained global optimal solution as the initial weight of the BP neural network, and training the BP neural network model; Step 3: Optimize the constructed passenger flow prediction model, correct the short-term passenger flow of the new line station obtained in step 2.4, and distribute the passenger flow.

2. The method for predicting short-term passenger flow at a new line station of urban rail transit according to claim 1 is characterized by: The classification of POI data in step 2.3 is as follows: (1) Non-mandatory travel POI data: including catering services, auto repair services, sports and leisure, scenic spots, and accommodation services; (2) Mandatory travel POI data: including companies, scientific research institutes, government organizations, financial services, and various schools; (3) Residential POI data: commercial residences.

3. The method for predicting short-term passenger flow at a new line station of urban rail transit according to claim 1 is characterized by: The cleaning and preprocessing in step 2.1 includes eliminating data information of roads or stations that are inaccessible in the short term due to construction, and data information of roads or stations that have been abolished or are about to be abolished due to the opening of new lines.

Images