CN112465396B - Bus scheduling method and system based on station events along line - Google Patents

Abstract

The invention discloses a bus scheduling method and system based on a station event along a line. The invention captures the things which are about to happen in a period of time in the future based on the web crawler, meanwhile, the neural network model is utilized to realize the prediction of the pedestrian volume by combining with the prior knowledge in daily life, the scheduling is carried out according to the result of the prediction of the pedestrian volume, and the targeted advance scheduling is realized according to the result of the prediction of the pedestrian volume which is likely to appear in a period of time in the future.

Description

A bus dispatching method and system based on station events along the route

Technical Field

The present invention relates to the technical field of public transportation dispatching, and in particular to a public transportation dispatching method and system based on line station events.

Background Art

Public transportation (abbreviated as "bus") is an important part of meeting daily travel needs. Effectively dispatching the bus system, avoiding traffic paralysis, and improving residents' travel efficiency are of great social significance.

Currently, the bus dispatching methods that have been put into use include: (1) increasing the number of vehicles on each line during the morning and evening peak hours; (2) extending the operating hours, etc. However, the current dispatching methods have little effect. The reason is that they do not conduct individualized and adaptive dispatching based on the characteristics of each line. Simply increasing the number of buses may aggravate the occurrence of traffic paralysis. In addition, the above methods cannot accurately predict the size of passenger flow that may occur in the future, so targeted advance dispatching cannot be achieved.

Summary of the invention

The purpose of the present invention is to provide a bus dispatching method and system based on station events along the route, so as to accurately predict the size of passenger flow that may occur in the future and realize targeted advance dispatching.

To achieve the above object, the present invention provides the following solutions:

A bus dispatching method based on station events along the route, the bus dispatching method comprising the following steps:

Use web crawler technology and prior knowledge learning methods to obtain information data on bus dispatch related events that are expected to occur along and around each bus route in the future period;

Performing data cleaning and feature extraction on the bus dispatching related event information data to obtain bus dispatching related event feature data;

Input the characteristic data of bus dispatch related events into the trained neural network model for passenger flow prediction to obtain the passenger flow along each bus line in the future period;

The buses along each bus line are dispatched according to the passenger flow along each bus line in the future period.

Optionally, the web crawler technology and prior knowledge learning method are used to obtain the bus dispatch related event information data expected to occur along and around each bus line in the future period, specifically including:

Using web crawler technology to obtain the information data of non-periodic bus dispatch related events expected to occur from the websites of relevant agencies along and around each bus route;

The prior knowledge learning method is adopted to analyze the historical periodic bus scheduling related events, and determine the distribution pattern of periodic bus scheduling related events along each bus line; the future time period is compared with the distribution pattern to determine the periodic bus scheduling related event information data of the future time period.

Optionally, data cleaning and feature extraction are performed on the bus dispatching related event information data to obtain bus dispatching related event feature data, specifically including:

Performing missing value processing, format conversion, duplicate data removal processing and noise data removal processing on the bus dispatch related event information data to obtain cleaned bus dispatch related event information data;

Using the data fitting method, the cleaned bus dispatch related event information data is verified to obtain the verified bus dispatch related event information data;

The Laida criterion is used to process the data of public transportation dispatching related event information that has passed the verification, and the public transportation dispatching related event information data after the data outlier processing is obtained;

Performing denoising and smoothing processing on the bus dispatching related event information data after outlier processing to obtain the bus dispatching related event information data after denoising and smoothing processing;

Segment the bus dispatch related event information data after denoising and smoothing to obtain multiple data segments;

Perform feature extraction on each data segment to obtain feature data of each data segment;

The characteristic data of each data segment are spliced together to obtain the characteristic data of bus scheduling related events.

Optionally, the neural network model is a decaying least squares model.

Optionally, the dispatching of buses along each bus line according to the passenger flow along each bus line in the future period specifically includes:

计算各个公交线路沿线的未来时段内的最大断面客流量；其中，p_i表示未来第i个公交线路沿线的未来时段内的最大断面客流量，N_l表示第i个公交线路沿线的人流量中的第l个站台的上车人数，Y_l表示第i个公交线路沿线的人流量中的第l个站台的下车人数，M_i表示第i个公交线路沿线的的站台数量，L表示第i个公交线路沿线的前L个站台；According to the passenger flow along each bus line in the future period, the formula

Calculate the maximum cross-sectional passenger flow in the future time period along each bus line; where _pi represents the maximum cross-sectional passenger flow in the future time period along the ith bus line, _Nl represents the number of passengers boarding the ith platform in the passenger flow along the ith bus line, _Yl represents the number of passengers getting off the ith platform in the passenger flow along the ith bus line, _Mi represents the number of platforms along the ith bus line, and L represents the first L platforms along the ith bus line;

计算各个公交线路沿线的未来时段内的最小车辆数；其中，n_i表示第i个公交线路沿线的未来时段内的最小车辆数。According to the maximum cross-sectional passenger flow in the future period along each bus line, the formula

Calculate the minimum number of vehicles along each bus line in the future period; where _ni represents the minimum number of vehicles along the ith bus line in the future period.

A bus dispatching system based on station events along the route, the bus dispatching system comprising:

The knowledge acquisition module is used to use web crawler technology and prior knowledge learning methods to obtain information data on bus dispatch related events that are expected to occur along and around various bus routes in the future period;

A data cleaning and feature extraction module, used to perform data cleaning and feature extraction on the bus dispatch related event information data to obtain bus dispatch related event feature data;

The passenger flow prediction module is used to input the characteristic data of bus scheduling related events into the trained neural network model for passenger flow prediction to obtain the passenger flow along each bus line in the future period;

The dispatching module is used to dispatch buses along each bus line according to the passenger flow along each bus line in the future period.

Optionally, the knowledge acquisition module specifically includes:

The web crawler submodule is used to obtain the information data of the non-periodic bus dispatch related events expected to occur along each bus route and the surrounding relevant institutions from the websites of the relevant institutions along each bus route and the surrounding relevant institutions by using the web crawler technology;

The prior knowledge learning submodule is used to adopt the prior knowledge learning method to analyze the historical periodic bus scheduling related events, determine the distribution pattern of periodic bus scheduling related events along each bus line; compare the future time period with the distribution pattern, and determine the periodic bus scheduling related event information data of the future time period.

Optionally, the data cleaning and feature extraction module specifically includes:

A data cleaning submodule is used to perform missing value processing, format conversion, duplicate data removal processing and noise data removal processing on the bus dispatch related event information data to obtain cleaned bus dispatch related event information data;

The data verification submodule is used to verify the cleaned bus dispatch related event information data by using the data fitting method to obtain the verified bus dispatch related event information data;

The data outlier processing submodule is used to process the data outliers of the verified bus dispatching related event information data using the Laida criterion to obtain the bus dispatching related event information data after outlier processing;

The denoising and smoothing processing submodule is used to perform denoising and smoothing processing on the bus dispatching related event information data after the outlier processing, so as to obtain the bus dispatching related event information data after the denoising and smoothing processing;

The data segmentation submodule is used to segment the bus dispatch related event information data after denoising and smoothing to obtain multiple data segments;

A feature extraction submodule is used to extract features from each data segment to obtain feature data of each data segment;

The data splicing submodule is used to splice the characteristic data of each data segment to obtain the characteristic data of bus scheduling related events.

Optionally, the neural network model is a decaying least squares model.

Optionally, the scheduling module specifically includes:

计算各个公交线路沿线的未来时段内的最大断面客流量；其中，p_i表示未来第i个公交线路沿线的未来时段内的最大断面客流量，N_l表示第i个公交线路沿线的人流量中的第l个站台的上车人数，Y_l表示第i个公交线路沿线的人流量中的第l个站台的下车人数，M_i表示第i个公交线路沿线的的站台数量，L表示第i个公交线路沿线的前L个站台；The maximum cross-section passenger flow calculation submodule is used to calculate the passenger flow along each bus line in the future period using the formula

Calculate the maximum cross-sectional passenger flow in the future time period along each bus line; where _pi represents the maximum cross-sectional passenger flow in the future time period along the ith bus line, _Nl represents the number of passengers boarding the ith platform in the passenger flow along the ith bus line, _Yl represents the number of passengers getting off the ith platform in the passenger flow along the ith bus line, _Mi represents the number of platforms along the ith bus line, and L represents the first L platforms along the ith bus line;

计算各个公交线路沿线的未来时段内的最小车辆数；其中，n_i表示第i个公交线路沿线的未来时段内的最小车辆数。The minimum vehicle number calculation submodule is used to calculate the minimum vehicle number according to the maximum cross-sectional passenger flow in the future period along each bus line using the formula

Calculate the minimum number of vehicles along each bus line in the future period; where _ni represents the minimum number of vehicles along the ith bus line in the future period.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

The present invention discloses a bus dispatching method and system based on line site events. The bus dispatching method: uses web crawler technology and prior knowledge learning methods to obtain bus dispatching related event information data that is expected to occur along each bus line and surrounding related institutions in the future period; performs data cleaning and feature extraction on the bus dispatching related event information data to obtain bus dispatching related event feature data; inputs the bus dispatching related event feature data into a trained neural network model for passenger flow prediction to obtain passenger flow along each bus line in the future period; dispatches buses along each bus line according to passenger flow along each bus line in the future period. The present invention is based on the web crawler to capture events that are about to occur in the future period, and at the same time combines daily prior knowledge, uses a neural network model to predict passenger flow, and dispatches according to the passenger flow prediction results, so as to achieve targeted advance dispatch according to the passenger flow size prediction results that may occur in the future period.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

FIG1 is a flow chart of a bus dispatching method based on station events along the route provided by the present invention;

FIG2 is a structural diagram of a bus dispatching system based on station events along the route provided by the present invention;

FIG3 is a bus stop map along bus route 870 in Taiyuan, Shanxi Province provided in Example 1 of the present invention;

FIG4 is a diagram showing the predicted results of the flow of people during the peak time period of the National Complex Network Conference provided by Example 1 of the present invention;

FIG5 is a diagram showing the result of the prediction of the flow of people during the peak time period of the National Computer Conference provided by Example 2 of the present invention;

FIG. 6 is a diagram showing the predicted results of the flow of people during the peak admission period of the Shanxi Sports Center provided by Example 2 of the present invention.

DETAILED DESCRIPTION

The purpose of the present invention is to provide a bus dispatching method and system based on station events along the route, so as to accurately predict the size of passenger flow that may occur in the future and realize targeted advance dispatching.

In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

When selecting bus stops, bus routes will give priority to institutions with relatively high population density or personnel flow along the route, such as stations, hospitals, schools, conference centers and hotels. These institutions usually have a series of regular events, such as: the passenger flow at stations is usually higher than usual during weekends and holidays; the passenger flow at school stops will also increase significantly during the school season; when there are large conferences (academic conferences, concerts, sports events, etc.), the passenger flow at conference centers and surrounding hotel stops will also increase significantly.

Therefore, we can effectively utilize the prior knowledge of possible events that may occur at bus stops and surrounding institutions to design an adaptive bus operation scheduling strategy.

As shown in FIG1 , the present invention provides a bus dispatching method based on station events along the route, and the bus dispatching method comprises the following steps:

Step 101, using web crawler technology and prior knowledge learning methods to obtain information data on bus scheduling related events that are expected to occur along and around various bus routes in the future period.

Step 101 uses web crawler technology and prior knowledge learning methods to obtain information data on bus scheduling related events that are expected to occur in the future period for relevant agencies along and around each bus line, specifically including: using web crawler technology to obtain information data on non-periodic bus scheduling related events that are expected to occur for relevant agencies along and around each bus line from the websites of relevant agencies along and around each bus line; using prior knowledge learning methods to analyze historical periodic bus scheduling related events to determine the distribution pattern of periodic bus scheduling related events along each bus line; comparing the future period with the distribution pattern to determine the periodic bus scheduling related event information data for the future period.

The surrounding related institutions include bus stations, hospitals, schools, conference centers and hotels along the bus routes.

Step 102: performing data cleaning and feature extraction on the bus dispatching related event information data to obtain bus dispatching related event feature data.

The step 102 of performing data cleaning and feature extraction on the bus dispatching related event information data to obtain bus dispatching related event feature data specifically includes: performing missing value processing, format conversion, duplicate data removal processing and noise data removal processing on the bus dispatching related event information data to obtain cleaned bus dispatching related event information data; using a data fitting method to verify the cleaned bus dispatching related event information data to obtain verified bus dispatching related event information data; using the Laida criterion to perform data outlier processing on the verified bus dispatching related event information data to obtain outlier-processed bus dispatching related event information data; performing denoising and smoothing processing on the bus dispatching related event information data after outlier processing to obtain denoising and smoothing-processed bus dispatching related event information data; performing data segmentation on the bus dispatching related event information data after denoising and smoothing to obtain multiple data segments; performing feature extraction on each data segment (i.e., performing missing value processing, format conversion, duplicate data removal processing and noise data removal processing on each data segment) to obtain feature data of each data segment; and splicing the feature data of each data segment to obtain bus dispatching related event feature data.

Step 103, inputting the characteristic data of bus scheduling related events into the trained neural network model for passenger flow prediction to obtain the passenger flow along each bus line in the future period.

The neural network model is a decaying least squares model.

Step 104, dispatching buses along each bus line according to the passenger flow along each bus line in the future time period.

The minimum number of buses can be derived when the full load rate is 120%. The maximum cross-sectional passenger flow on all buses in each time period is calculated based on the number of people getting on and off the buses, and the minimum number of buses required in this time period is calculated based on the maximum number of people. Assuming that passengers in each time period arrive at the platform evenly, in order to ensure that every passenger in each time period can board the bus, the upper limit of the capacity of each bus is 120 people.

计算各个公交线路沿线的未来时段内的最大断面客流量；其中，p_i表示未来第i个公交线路沿线的未来时段内的最大断面客流量，N_l表示第i个公交线路沿线的人流量中的第l个站台的上车人数，Y_l表示第i个公交线路沿线的人流量中的第l个站台的下车人数，M_i表示第i个公交线路沿线的的站台数量，L表示第i个公交线路沿线的前L个站台；根据各个公交线路沿线的未来时段内的最大断面客流量，利用公式

计算各个公交线路沿线的未来时段内的最小车辆数；其中，n_i表示第i个公交线路沿线的未来时段内的最小车辆数。Step 104 is to dispatch buses along each bus line according to the passenger flow along each bus line in the future time period, specifically including: according to the passenger flow along each bus line in the future time period, using the formula

Calculate the maximum cross-sectional passenger flow in the future time period along each bus line; where _pi represents the maximum cross-sectional passenger flow in the future time period along the i-th bus line, _Nl represents the number of passengers boarding the l-th platform in the passenger flow along the i-th bus line, _Yl represents the number of passengers getting off the l-th platform in the passenger flow along the i-th bus line, _Mi represents the number of platforms along the i-th bus line, and L represents the first L platforms along the i-th bus line; according to the maximum cross-sectional passenger flow in the future time period along each bus line, use the formula

Calculate the minimum number of vehicles along each bus line in the future period; where _ni represents the minimum number of vehicles along the ith bus line in the future period.

As shown in FIG2 , the present invention further provides a bus dispatching system based on station events along the route, the bus dispatching system comprising:

The knowledge acquisition module is used to use web crawler technology and prior knowledge learning methods to obtain information data on bus scheduling related events that are expected to occur along each bus route and surrounding related agencies in the future period.

The knowledge acquisition module specifically includes: a web crawler submodule, which is used to use web crawler technology to obtain the expected non-periodic bus scheduling related event information data of the relevant institutions along and around each bus line from the websites of the relevant institutions along and around each bus line; a priori knowledge learning submodule, which is used to use a priori knowledge learning method to analyze the historical periodic bus scheduling related events and determine the distribution pattern of the periodic bus scheduling related events along each bus line; compare the future time period with the distribution pattern to determine the periodic bus scheduling related event information data of the future time period.

The main function of the knowledge acquisition module is to use multiple channels to search for information related to bus routes and stops.

First, the web crawler module is used to take the various stations along the bus line (such as stations, schools, hospitals, conference centers, hotels, etc.) as search terms, and through various channels (such as official websites, news portals, Weibo and WeChat public accounts, etc.), search for related events that will occur at each station along the line in the future (such as Spring Festival travel, the start of the school season, large-scale conferences, performances, etc.).

Secondly, use the prior knowledge learning submodule to predict related events that will occur in the future (for example, weekends and holidays, travel patterns, etc.). The specific method is to analyze the frequency spectrum of the event through the FFT algorithm, obtain the distribution law of the relevant data of the flow of people at each bus stop, and then perform the maximum overlapping discrete wavelet decomposition of the time series, integrate the data to establish the ARIMA model, and compare the future time points with the time series of historical events. Use the Fourier spectrum to analyze the probability of the event, and then predict the related events that will occur in the future. For example, during holidays or commuting time, the model can be used to predict the law of human flow with prior knowledge.

The data cleaning and feature extraction module is used to perform data cleaning and feature extraction on the bus dispatch related event information data to obtain bus dispatch related event feature data.

The main function of the data cleaning and feature extraction module is to clean and extract features from the data retrieved by the knowledge acquisition module, analyze the obtained spectrum results using FFT software, remove the interference data that deviates from the fitting curve, and obtain event information and features that are truly related to the stations along the bus line. For example: the address of the meeting, the scale of the meeting, the distribution of participants, the peak event period of travel (entry and exit), the first and last bus times, etc. Among them, the steps of data cleaning are: 1) missing value processing, according to the missing rate and importance of the event, it is divided into removing fields, filling missing values, and re-obtaining data; 2) format and content processing; 3) removing duplicate data; 4) processing of noise data. The steps of data feature extraction are: 1) data authenticity judgment, using fitting technology and other means to verify the authenticity of the data; 2) data outlier processing, using data outlier processing based on the Laida criterion; 3) data denoising and smoothing processing; 4) data segmentation; 5) feature data extraction.

Specifically, the data cleaning and feature extraction module specifically includes: a data cleaning submodule, which is used to perform missing value processing, format conversion, duplicate data removal processing and noise data removal processing on the bus scheduling related event information data to obtain the cleaned bus scheduling related event information data; a data verification submodule, which is used to verify the cleaned bus scheduling related event information data using a data fitting method to obtain the verified bus scheduling related event information data; a data outlier processing submodule, which is used to perform data outlier processing on the verified bus scheduling related event information data using the Laida criterion to obtain the bus scheduling related event information data after outlier processing; a denoising and smoothing processing submodule, which is used to perform denoising and smoothing processing on the bus scheduling related event information data after outlier processing to obtain the bus scheduling related event information data after denoising and smoothing processing; a data segmentation submodule, which is used to perform data segmentation on the bus scheduling related event information data after denoising and smoothing processing to obtain multiple data segments; a feature extraction submodule, which is used to perform feature extraction on each data segment to obtain feature data of each data segment; a data splicing submodule, which is used to splice the feature data of each data segment to obtain the feature data of bus scheduling related events.

The passenger flow prediction module is used to input the characteristic data of bus scheduling related events into the trained neural network model for passenger flow prediction, and obtain the passenger flow along each bus line in the future period.

The main function of the passenger flow prediction module is to predict the size of the passenger flow with travel demand during the time period when the event occurs based on the valid data collected, cleaned and extracted by the data cleaning and feature extraction module. In the process of bus scheduling, since the trend of passenger flow change is highly random and has obvious nonlinear characteristics, it is not suitable to use traditional prediction methods to predict passenger flow. Therefore, the present invention adopts a data mining method, an artificial neural network system, and uses historical data for modeling and prediction. Artificial neural networks can achieve various linear and nonlinear mapping capabilities by changing the type of excitation function. At the same time, the neural network model has strong learning and adaptive capabilities, so it can adapt to the characteristics of short-term passenger flow to a large extent, thereby ensuring the reliability of prediction. The present invention selects the attenuated least squares algorithm as the algorithm of the artificial neural network. The first step in establishing the prediction model is to obtain data. Therefore, based on the valid data obtained in module 2, the data corresponding to two adjacent weeks, two adjacent days, and two adjacent time periods are selected as input samples, and the artificial neural network model is used to predict passenger flow. In order to verify the correctness of the prediction model, the data samples can be divided into a training set and a test set, and the ratio is 3:1. Before using the neural network model for prediction, the data needs to be normalized to between [-1,1] to simplify the prediction and prevent the problem of neuron output saturation. Once the preliminary work is completed, the prediction model based on the artificial neural network can be written through simulation software such as MATLAB, and then the training parameters of the network are set, the network is initialized, and then the prediction network is trained, the training samples are input, and finally the trained network is tested.

The dispatching module is used to dispatch buses along each bus line according to the passenger flow along each bus line in the future period.

The main function of the scheduling module is to establish an optimized bus scheduling model based on the passenger flow prediction model constructed by the passenger flow prediction module. The present invention uses a web crawler to capture the data of passenger flow and bus schedules in the past period of time, and cleans and extracts the data. Combined with the actual situation, according to the characteristics of each bus line station, a structural equation model is constructed. For example, a minimum bus model can be established: the mathematical expression of the model is as follows:

The minimum value of vehicles in this time period is:

p _i represents the maximum cross-sectional passenger flow along the i-th bus line in the future time period, N _l represents the number of passengers getting on the l-th platform along the i-th bus line, Y _l represents the number of passengers getting off the l-th platform along the i-th bus line, M _i represents the number of platforms along the i-th bus line, and L represents the first L platforms along the i-th bus line.

Structural equation model is a statistical method based on variables to analyze the relationship between variables. To establish the model, the common parameters are first established and calculated. The influencing factors (peak hours, special time points, road conditions, weather, passenger comfort, etc.) are brought into the model for quantitative calculation to obtain the minimum number of buses required, and optimization strategies and suggestions are proposed based on the model results. For example, taking No. 1 bus in Taiyuan, Shanxi as an example, the bus route, departure schedule, station distance table and other data are established to construct the above model. By analyzing the model results and combining them with reality, the current bus route scheduling method is adjusted and optimized. By adjusting 1) whether it is necessary to add more buses, 2) if it is necessary to add more buses, how many buses need to be added in which time period, the scheduling arrangement for a period of time in the future is formulated.

计算各个公交线路沿线的未来时段内的最大断面客流量；其中，p_i表示未来第i个公交线路沿线的未来时段内的最大断面客流量，N_l表示第i个公交线路沿线的人流量中的第l个站台的上车人数，Y_l表示第i个公交线路沿线的人流量中的第l个站台的下车人数，M_i表示第i个公交线路沿线的的站台数量，L表示第i个公交线路沿线的前L个站台；最小车辆数计算子模块，用于根据各个公交线路沿线的未来时段内的最大断面客流量，利用公式

计算各个公交线路沿线的未来时段内的最小车辆数；其中，n_i表示第i个公交线路沿线的未来时段内的最小车辆数。Specifically, the scheduling module specifically includes: according to the passenger flow along each bus line in the future period, using the formula

Calculate the maximum cross-sectional passenger flow in the future time period along each bus line; where _pi represents the maximum cross-sectional passenger flow in the future time period along the i-th bus line, _Nl represents the number of passengers boarding the l-th platform in the passenger flow along the i-th bus line, _Yl represents the number of passengers getting off the l-th platform in the passenger flow along the i-th bus line, _Mi represents the number of platforms along the i-th bus line, and L represents the first L platforms along the i-th bus line; the minimum vehicle number calculation submodule is used to calculate the maximum cross-sectional passenger flow in the future time period along each bus line using the formula

Calculate the minimum number of vehicles along each bus line in the future period; where _ni represents the minimum number of vehicles along the ith bus line in the future period.

The technical solution of the present invention is described below in conjunction with two specific embodiments.

Embodiment 1:

This embodiment takes bus route 870 in Taiyuan, Shanxi Province as the research object, and uses the method proposed in the present invention to study the dispatching strategy of this route. As shown in Figure 3, bus route 870 includes 26 stops, including typical stops with large passenger flow, such as: railway station, schools (Shanxi University, Judicial School and University of Finance and Economics, etc.), companies (Foxconn), hospitals (Shanxi Provincial Hospital of Traditional Chinese Medicine and Shanxi Provincial People's Hospital are located near the Bingzhou East Street Station on Bingzhou North Road), and conference centers/hotels (Shanxi International Conference Center and multiple hotels are located near the Wuyi Square Station).

Taking September 2016 as an example, the web crawler module searched for information related to bus stops along Route 870 in October 2016. The upcoming events are two national academic conferences to be held on October 14-17 (National Complex Network Conference) and October 20-22 (National Computer Conference) at Longcheng International Hotel and Lakeside International Hotel (main venue, including 5 branch venues), both of which are near the May Day Square station. At the same time, the information searched by the web crawler module shows that: (1) the organizer of the above two conferences is Shanxi University; (2) the scale of the National Complex Network Conference is about 1,000 people; the scale of the National Computer Conference is about 5,000 people. Combined with daily prior knowledge, it can be known that during the conference, a large number of teachers and students will travel between the venue and the school. Shanxi University is also along Route 870. Therefore, it can be inferred that the passenger flow of Route 870 during the conference will be significantly higher than usual.

In addition, according to the meeting schedule searched by the web crawler module, it can be seen that the meetings all start at 8:30 in the morning and end at 18:00 in the afternoon. According to the prior knowledge module, participants generally arrive at the venue within half an hour before the meeting, that is, the peak time period for admission in the morning is 8:00-8:30. Without loss of generality, it is assumed that the flow of people during the peak time period satisfies the normal distribution, and the number of participants traveling between Shanxi University and May Day Square during the two meetings in the morning accounts for 20% of the total number of people (200 and 1000 people respectively). According to the crowd flow prediction module, the number of people traveling at each time point during the peak time period can be obtained, as shown in Figures 4 and 5 respectively.

In addition, considering that there are other bus routes (such as 103, 812, etc.) between Shanxi University and May Day Square, based on the above analysis, the bus dispatching strategy for route 870 during the two conferences can be formulated, as shown in Table 1.

Table 1 The bus dispatching strategy for route 870 to meet the travel needs of participants during the two conferences

It can be seen intuitively from Table 1 that during the National Complex Network Conference, there is basically no need to dispatch more buses because the scale of the conference is relatively small; during the National Computer Conference, additional buses need to be added according to the law of approximate normal distribution to meet the travel needs of participants. It should be noted that the time period in Table 1 refers to the time to arrive at the Wuyi Square station, and each bus line needs to formulate a departure strategy in advance according to the full time of the line.

Embodiment 2:

This embodiment takes Shanxi Sports Center as the research object, and uses the method proposed in the present invention to study the bus dispatching strategy of the site. As is known to all, Shanxi Sports Center is also the home court of Shanxi Fenjiu Men's Basketball Team. During the CBA season (from October to February of the following year), there are intensive competitions. Taking November 2016 as an example, the home schedule of Shanxi Fenjiu Men's Basketball Team searched by the web crawler module is listed in Table 2. As can be seen from the table, the start time of all 6 games is 19:35.

At the same time, combined with the following two prior knowledge: (1) the general public will enter the stadium 30-60 minutes in advance, so it can be known that the peak time for the audience to enter the stadium is 18:30-19:30; (2) under normal circumstances (excluding overtime), a CBA game (including halftime) lasts about 2 hours, so it can be known that the peak time for the audience to leave the stadium is 21:30-22:00.

Table 2 CBA 2016-2017 season Shanxi Fenjiu men's basketball home schedule (November 2016)

Round Matchup (home team first) time 3 Shanxi vs Shenzhen 11-04 19:35 4 Shanxi vs Tianjin 11-06 19:35 6 Shanxi vs Beijing 11-13 19:35 7 Shanxi vs Liaoning 11-16 19:35 11 Shanxi vs Fujian 11-27 19:35 12 Shanxi vs Zhejiang 11-30 19:35

The bus routes around Shanxi Sports Center searched by the web crawler module are listed in Table 3. As can be seen from the table, all routes end their operations before 20:00, so they cannot meet the travel needs of the audience leaving the venue.

Table 3 Bus routes around Shanxi Sports Center

At the same time, the web crawler module searched for information about the attendance rate of Shanxi Fenjiu Men's Basketball Team. After data cleaning and feature analysis, it showed that the total number of spectators that can be accommodated is 8,000, and the average attendance rate is about 70%, that is, there are about 5,600 spectators per game. Conservatively assuming that about 60% of the spectators choose public transportation, it can be inferred that there are about 3,360 people per game who have travel needs during the two peak event periods of entrance and exit.

In addition, without loss of generality, assuming that during the two peak time periods of entry and exit, the flow of people satisfies an approximate normal distribution, the flow of people at each time point can be obtained using the flow prediction model. Here, the interval is 5 minutes, and the results are shown in Figure 6.

Therefore, based on the above analysis, a bus dispatching strategy that meets the admission demand of the Shanxi Fenjiu Men's Basketball Team can be formulated (here, it is assumed that the number of spectators from the five bus routes in Table 3 is equal, that is, the dispatching strategies of the five routes are the same), as shown in Table 4, that is, the bus dispatching strategy also approximately satisfies the normal distribution. It should be noted that the time period in Table 4 refers to the time of arrival at the Shanxi Sports Center, and each bus route needs to formulate a departure strategy in advance according to the full journey time of the route.

Table 4 Bus dispatch strategy to meet the admission demand of Shanxi Fenjiu Men's Basketball Team spectators

Time period The number of vehicles dispatched on each route during this time period 18:30–18:40 1 18:40–18:55 2 18:55–19:05 3 19:05–19:20 2 19:20–19:35 1

The bus dispatching strategy during the peak exit period is similar to that during the peak entrance period. Compared with the current operation plan, the dispatching strategy provided in this embodiment is mainly reflected in two aspects: (1) adding evening buses (21:30-22:00) on home game days; (2) adaptively adjusting the frequency and interval of shuttle buses during the peak entrance and exit periods according to an approximate normal distribution.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

The present invention discloses a bus dispatching method and system based on line site events. The bus dispatching method: uses web crawler technology and prior knowledge learning methods to obtain bus dispatching related event information data that is expected to occur along each bus line and surrounding related institutions in the future period; performs data cleaning and feature extraction on the bus dispatching related event information data to obtain bus dispatching related event feature data; inputs the bus dispatching related event feature data into a trained neural network model for passenger flow prediction to obtain passenger flow along each bus line in the future period; dispatches buses along each bus line according to passenger flow along each bus line in the future period. The present invention is based on the web crawler to capture events that are about to occur in the future period, and at the same time combines daily prior knowledge, uses a neural network model to predict passenger flow, and dispatches according to the passenger flow prediction results, so as to achieve targeted advance dispatch according to the passenger flow size prediction results that may occur in the future period.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

This article uses specific examples to illustrate the principles and implementation methods of the invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. The described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

Claims (6)

1. A bus scheduling method based on along-line station events is characterized by comprising the following steps:

acquiring predicted bus scheduling related event information data of related mechanisms along each bus route and around the bus route in a future period by adopting a web crawler technology and a priori knowledge learning method;

the method for acquiring the predicted bus scheduling related event information data of the related mechanisms along each bus route and around in the future period by adopting the web crawler technology and the priori knowledge learning method specifically comprises the following steps:

acquiring expected aperiodic bus dispatching related event information data of the related mechanisms along each bus route and around the bus route from websites of the related mechanisms along each bus route and around the bus route by adopting a web crawler technology;

analyzing historical periodic bus scheduling related events by adopting a priori knowledge learning method, and determining the distribution rule of the periodic bus scheduling related events along each bus route; comparing the future time period with the distribution rule, and determining the periodic bus scheduling related event information data of the future time period;

performing data cleaning and feature extraction on the bus scheduling related event information data to obtain bus scheduling related event feature data;

carrying out data cleaning and feature extraction on the public transportation scheduling related event information data to obtain public transportation scheduling related event feature data, and specifically comprising the following steps:

carrying out missing value processing, format conversion, repeated data removal processing and noise data removal processing on the bus scheduling related event information data to obtain cleaned bus scheduling related event information data;

verifying the cleaned bus scheduling related event information data by using a data fitting method to obtain verified bus scheduling related event information data;

carrying out data abnormal value processing on the public transport scheduling related event information data passing the verification by adopting a Lauda criterion to obtain the public transport scheduling related event information data after the abnormal value processing;

denoising and smoothing the public transportation scheduling related event information data after abnormal value processing to obtain the public transportation scheduling related event information data after denoising and smoothing;

carrying out data segmentation on the bus scheduling related event information data subjected to denoising and smoothing to obtain a plurality of data segments;

extracting the characteristics of each data segment to obtain the characteristic data of each data segment;

splicing the characteristic data of each data segment to obtain the characteristic data of the bus dispatching related events;

inputting the characteristic data of the bus scheduling related events into a trained neural network model for traffic prediction to obtain the traffic along each bus line in a future period;

and dispatching the buses along each bus line according to the pedestrian volume along each bus line in the future period.

2. The method of claim 1, wherein the neural network model is a least squares model of attenuation.

3. The bus scheduling method based on the bus stop events along the line as claimed in claim 1, wherein the bus scheduling method for the bus along each bus line according to the pedestrian volume along each bus line in the future time period specifically comprises:

according to the pedestrian volume along each bus line in the future period, a formula is utilized

Calculating the maximum section passenger flow in the future time period along each bus line; wherein p is _i Represents the maximum section passenger flow in the future time interval along the ith future bus route, N _l The number of passengers getting on the bus at the first station in the passenger flow along the ith bus line is shown, Y _l The number of people getting off the first platform in the flow of people along the ith bus line is represented, M _i The number of the stations along the ith bus line is represented, and L represents the front L stations along the ith bus line;

according to the maximum section passenger flow in the future time interval along each bus line, the formula is utilized

Calculating the future time interval along each bus lineMinimum number of vehicles in; wherein n is _i Representing the minimum number of vehicles in a future period along the ith bus route.

4. A bus dispatching system based on station events along the line is characterized by comprising:

the knowledge acquisition module is used for acquiring the information data of the predicted bus dispatching related events of the related mechanisms along each bus route and around the bus route in the future period by adopting a web crawler technology and a priori knowledge learning method;

the knowledge acquisition module specifically comprises:

the network crawler submodule is used for acquiring the predicted aperiodic bus scheduling related event information data of the related mechanisms along each bus route and around from the websites of the related mechanisms along each bus route and around by adopting the network crawler technology;

the prior knowledge learning submodule is used for analyzing historical periodical bus scheduling related events by adopting a prior knowledge learning method and determining the distribution rule of the periodical bus scheduling related events along each bus route; comparing the future time period with the distribution rule, and determining the periodic bus scheduling related event information data of the future time period;

the data cleaning and feature extraction module is used for cleaning and extracting the data of the event information related to the bus dispatching to obtain the feature data of the event related to the bus dispatching;

the data cleaning and feature extraction module specifically comprises:

the data cleaning submodule is used for carrying out missing value processing, format conversion, repeated data removal processing and noise data removal processing on the public transportation scheduling related event information data to obtain cleaned public transportation scheduling related event information data;

the data verification submodule is used for verifying the cleaned bus dispatching related event information data by using a data fitting method to obtain the verified bus dispatching related event information data;

the data abnormal value processing submodule is used for carrying out data abnormal value processing on the verified bus dispatching related event information data by adopting a Lauda criterion to obtain the bus dispatching related event information data after the abnormal value processing;

the denoising and smoothing sub-module is used for denoising and smoothing the bus scheduling related event information data after the abnormal value processing to obtain the bus scheduling related event information data after the denoising and smoothing processing;

the data segmentation submodule is used for carrying out data segmentation on the bus scheduling related event information data subjected to denoising and smoothing processing to obtain a plurality of data segments;

the characteristic extraction submodule is used for extracting the characteristics of each data segment to obtain the characteristic data of each data segment;

the data splicing submodule is used for splicing the characteristic data of each data segment to obtain the characteristic data of the bus dispatching related events;

the passenger flow prediction module is used for inputting the characteristic data of the bus scheduling related events into a trained neural network model for passenger flow prediction to obtain the passenger flow along each bus line in a future period;

and the scheduling module is used for scheduling the buses along each bus line according to the pedestrian volume along each bus line in a future time period.

5. The bus dispatching system based on events along line stops of claim 4, wherein the neural network model is a least squares model of attenuation.

6. The bus scheduling system based on events along line stops as claimed in claim 4, wherein the scheduling module specifically comprises:

a maximum section passenger flow volume calculation submodule for utilizing a formula according to the passenger flow volume along each bus line in a future time period

Calculating the maximum section passenger flow in the future time period along each bus line; wherein p is _i Represents the maximum section passenger flow quantity N in the future time interval along the ith bus route _l The number of passengers getting on the bus at the first station in the passenger flow along the ith bus line is shown, Y _l The number of people getting off at the first platform in the pedestrian flow along the ith bus line is represented, M _i The number of the stations along the ith bus line is represented, and L represents the front L stations along the ith bus line;

a minimum vehicle number calculation submodule for utilizing a formula according to the maximum section passenger flow in the future time period along each bus line

Calculating the minimum number of vehicles in a future time period along each bus line; wherein n is _i Representing the minimum number of vehicles in a future period along the ith bus route. />

Images