A trip OD simulation method based on boarding volume at public transport stations

Abstract

The application discloses a kind of travel OD simulation methods based on public transport site boarding volume, it is related to traffic planning technical field, the simulation method includes the following steps: to the corresponding boarding record of each time of user boarding process Generation and the OD matrix of user;The distribution of station is analyzed, and the boarding and alighting frequency of each station is obtained;The multidimensional information recorded in each boarding record of any selected station is analyzed, and the association of dimension information and boarding and alighting frequency is obtained;The influence of any dimension information is analyzed, and different dimension information is handled in layers;Each dimension information is allocated corresponding characteristic coefficient, and the boarding and alighting simulation model is constructed to simulate the boarding and alighting of each station;Real-time acquisition of the multidimensional information of each station, based on boarding and alighting simulation model, the expected boarding and alighting is predicted, and abnormal identification is carried out;Effectively reduce the error generated by OD simulation prediction, improve accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of traffic planning, and particularly relates to a travel OD simulation method based on public transport site boarding volume. BACKGROUND

[0002] An OD matrix is a core tool for urban traffic planning and management, which analyzes the distribution of passenger travel origins and destinations to provide a basis for bus scheduling, road network optimization, etc. Traditional OD simulation methods mainly rely on static data such as historical survey data and fixed-period boarding volume statistics to infer travel demand through probability models or gravity models.

[0003] However, with the increasing complexity of urban traffic, the limitations of static models gradually appear: traditional methods cannot reflect the spatiotemporal changes of traffic flow in real time, such as peak tidal passenger flow and abnormal fluctuations caused by sudden events, leading to prediction bias; and most methods only rely on a single data source, lack of fusion with multi-source data such as weather and social holidays, and are difficult to capture the influence of external factors on OD distribution. SUMMARY

[0004] The present application aims to provide a travel OD simulation method based on public transport site boarding volume to solve the problems in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a travel OD simulation method based on public transport site boarding volume, the simulation method comprising the following steps:

[0006] Step S100: record the sites passed by the user in each ride process after user authorization, collect multi-dimensional information of each site in different ways during the ride process, generate corresponding ride records, and generate the OD matrix of the user based on all ride records of the user;

[0007] Step S200: analyze the distribution of all sites based on the OD matrix of each user to obtain the boarding and alighting frequency of each site; select an arbitrary site, analyze the multi-dimensional information recorded in each ride record of the selected site, and obtain the association between different dimensional information and the boarding and alighting frequency;

[0008] Step S300: based on the association of different dimensional information, analyze the influence of arbitrary dimensional information, and set hierarchical judgment conditions to process different dimensional information hierarchically;

[0009] Step S400: according to the influence of different dimensional information and the hierarchical situation, assign corresponding characteristic coefficients to each dimensional information;

[0010] Step S500: Constructing a boarding and alighting simulation model to simulate boarding and alighting conditions at each station; training and correcting the boarding and alighting simulation model based on the multi-dimensional information of each ride record;

[0011] Step S600: acquiring multi-dimensional information of each current station in real time, predicting the expected landing and boarding volume of each station based on the landing and boarding simulation model, and identifying abnormalities at each station;

[0012] Step S700: Analyze the actual boarding and landing frequencies of each abnormal station after the scheduling strategy is adjusted, perform abnormal analysis on the adjustment of the scheduling strategy, and provide feedback.

[0013] Furthermore, step S100 includes the following steps:

[0014] Step S101: Set a boarding quota and a disembarkation quota for each station. Whenever a user boards public transportation at a station, the boarding quota for the station is incremented by one. Each station the user passes through is recorded in real time during the ride. When the user disembarks at another station, the disembarkation quota for the other station is incremented by one.

[0015] Step S102: setting a station where the user boards public transportation as the initial station and another station where the user gets off as the terminal station, sorting all stations that the user passes through from the initial station to the terminal station in order, and generating a travel route for the user; selecting any station from the travel route, and adopting several preset information collection methods to obtain dimensional information of the selected station in several dimensions, wherein one information collection method corresponds to dimensional information of one dimension, and dimensional information of any station in various dimensions is obtained; the several information collection methods include directly obtaining through the network whether the current time point is a travel peak period or a holiday, etc., using environmental sensors to monitor the rainfall and temperature of the environment where the station is located, and judging whether there are any emergencies such as construction or activities nearby by collecting network information, and obtaining dimensional information of various dimensions such as time dimension, environmental dimension, and event dimension;

[0016] Step S103: Summarize the bus routes including each station and the dimensional information of any station in each dimension to generate a bus record of the user; summarize all the bus records of the user to generate an OD matrix of the user.

[0017] Furthermore, step S200 includes the following steps:

[0018] Step S201: Obtain the OD matrix of all users, count the number of times each station is the initial station and the terminal station, and set the number of times the i-th station is the initial station to m i and the number of terminal stations is ni ; Count the total number of ride records of all users as Num, and calculate the average boarding and alighting frequency f at the i-th station i =(m i +n i ) / Num; the boarding and alighting frequency indicates the frequency of all users getting on and off the bus at the station;

[0019] Step S202: Obtain the travel time period of each travel record, preset a unit period T, arbitrarily select the j-th unit period, and obtain the time interval of the j-th unit period as TR jT =[(j-1)T,jT); arbitrarily select a travel time period (t1,t2) of a travel record, if (t1,t2)∈TR jT , the selected ride record is set as the target ride record for the jth unit cycle; because external factors include time and environment dimensions, both of which change with time in a unit cycle and produce certain patterns. For example, morning and evening rush hours or high temperatures at noon will affect users' travel. Therefore, dividing the unit cycle can facilitate subsequent comparison and analysis of the impact.

[0020] Step S203: Extract all target ride records in the jth unit cycle and obtain the boarding and alighting frequency (f i ) j , and obtain the deviation degree of the i-th station (η i ) j =[(f i ) j -f i ] / f i ; Randomly select a target ride record. If the i-th station is the starting station or the terminal station in the selected target ride record, obtain the dimension information of the i-th station in each dimension, extract features from the dimension information of each dimension, and obtain the correlation coefficient g of the extracted features in the j-th unit period j =(η i ) j The correlation between various external factors is directly related to the deviation of landing and landing frequencies. If a change in an external factor causes the deviation of landing and landing frequencies to increase, it means that the correlation between the two is high.

[0021] Step S204: arbitrarily select a dimension and perform similarity comparison on the features of the selected dimension in each target ride record. If the obtained similarity exceeds the preset similarity threshold, the compared feature is set as a similar feature, and the correlation coefficient of the compared feature is set as a similar correlation coefficient. The average value of the similar correlation coefficient is calculated to obtain the average correlation coefficient (g j ) aveAnd match with the corresponding same kind of feature, get the average correlation coefficient between different dimension information and the frequency of boarding and alighting in each dimension.

[0022] Further, step S300 includes the following steps:

[0023] Step S301: Arbitrarily select two unit periods, select a dimension of dimension information from each of the two unit periods, extract the same features between the two unit periods from the same dimension, and respectively to the average correlation coefficient corresponding to the same feature, to get the correlation coefficient difference Δg between the two unit periods;

[0024] Step S302: Obtain the corresponding dimension information of the same feature in the two unit periods, compare the dimension information of the two unit periods, obtain the deviation degree Δp of the same feature between the two unit periods, and calculate the influence degree y=Δg / Δp of the same feature; The deviation degree of the same feature directly affects the correlation between the boarding and alighting frequency, therefore, direct division of the two can directly reflect the influence degree of the feature; Because the change caused by the change of the feature is inversely affected, for example, the temperature will be higher and higher, and the number of trips will be less and less, therefore, the influence degree will exist in positive and negative situations;

[0025] Step S303: Preset an expected influence range (y min ,y max ), get the difference Δy ex =y max -y min , and divide the numerical range into a levels, then get the influence degree range of the bth level , if y∈R b , the dimension information of the extracted same feature is divided into the bth level.

[0026] Further, step S400 includes the following steps:

[0027] Step S401: Arbitrarily select a dimension of dimension information, obtain each feature under the selected dimension, get the level of each feature, and set the selected dimension to have c features, wherein the dth feature is in the b(d)th level.

[0028] Step S402: Preset a distributable feature value Z, distribute the feature value Z to each level according to the set proportion, according to the formula:

[0029] ;

[0030] Wherein, b is a positive integer and b∈(1,a); the distributable coefficient of the b(d)th level is calculated as X b(d); the default feature value Z is 1, the higher the level of the feature is, the greater the degree of influence is, and the higher the feature value allocated is;

[0031] Step S403: set the number of features contained in the b(d)th level as u b(d) , obtain the allocation coefficient of the dth feature in the selected dimension as X(d)=X b(d) / u b(d) , obtain the degree of influence of the dth feature as y(d), according to the formula:

[0032] ;

[0033] Calculate the feature coefficient V of the dimension information of the selected dimension; the feature coefficient is a comprehensive analysis of the degree of influence and the allocation coefficient of the feature, the degree of influence directly reflects the influence of the feature on the landing and take-off frequency, and the allocation coefficient reflects the degree of correlation between the feature and the landing and take-off frequency, which can indirectly reflect the degree of influence of the feature, and the combination of the two can more accurately obtain the feature coefficient of each dimension, that is, a comprehensive influence coefficient of each dimension on the user landing and take-off quantity. Since the influence effect of different situations under one dimension will exist differences, for example, higher or lower temperature will lead to a decrease in landing and take-off frequency, and the degree of influence will exist differences, therefore, it is necessary to comprehensively analyze different situations to obtain more accurate influence coefficients, which is beneficial to the accuracy of subsequent simulation model construction.

[0034] Further, step S500 includes the following steps:

[0035] Step S501: randomly select a station and a ride record containing the selected station, set the selected ride record as a test record, from all ride records containing the selected station, obtain a plurality of ride records whose ride time periods are located before the test record and are adjacent and continuous, obtain the dimension information of each dimension in the plurality of ride records, and obtain the data change interval of the same type of feature under any dimension;

[0036] Step S502: obtain the ride time periods of the plurality of ride records, merge to obtain a continuous time period, obtain the landing and take-off frequency f of the selected station in the continuous time period, obtain the dimension information of each dimension in the test record, and compare the deviation with the corresponding data change interval in the plurality of ride records, and calculate the deviation degree of each dimension;

[0037] Step S503: set the deviation degree of the kth dimension in the test record as q k , obtain the feature coefficient of the dimension information of the kth dimension as V k , and construct a landing and take-off simulation model:

[0038] ;

[0039] wherein c ’ is the number of dimensions contained in the test record, s is a constant coefficient; the actual boarding and alighting frequency of the test record in the ride time period is f ac , a preset allowed frequency deviation Δf th is obtained, the numerical range of the constant coefficient s is obtained, so that |f ’ -f ac |≤Δf th ;

[0040] Step S504: The remaining ride records are randomly selected to train the boarding and alighting simulation model, and the constant coefficient s is corrected. If the corrected constant coefficient s has several numerical values that satisfy |f ’ -f ac |≤Δf th , a numerical value is randomly selected to confirm the constant coefficient s. The boarding and alighting simulation model is based on the average boarding and alighting frequency in the recent period of time, adjusts the historical average boarding and alighting frequency according to the dimensional information of each dimension in the current time period and the change before that in combination with the influence degree of each dimension, and sets the constant coefficient to help the deviation of the output result of the model to be within an allowed range. The numerical value of the constant coefficient is continuously adjusted by the historical data to obtain a more accurate simulation model for effectively predicting the expected boarding volume of each station.

[0041] Further, step S600 includes the following steps:

[0042] Step S601: Obtain a plurality of unit periods with the shortest time interval from the current time, and obtain the boarding and alighting frequencies of each station in the plurality of unit periods and the data change intervals of each station in each dimension;

[0043] Step S602: Obtain the dimensional information of each station in each dimension at the current time, and obtain the deviation degree of each dimension. Arbitrarily select a station, input the boarding and alighting frequencies of the selected station in the plurality of unit periods and the deviation dimensions of each dimension into the boarding and alighting simulation model to obtain the expected boarding and alighting frequency f ex of the selected station.

[0044] Step S603: Set the average boarding and alighting frequency of the i-th station as f i , obtain the expected boarding and alighting frequency of the i-th station at the current time as (f ex ) i , and if |f i -(f ex ) i |≥Δf thIf the prediction result is greatly different from the historical average result, it indicates that the normal vehicle scheduling frequency cannot meet the actual demand of the station, and the waiting time of public transportation can be shortened or lengthened according to the demand.

[0045] Further, the step S700 comprises the following steps:

[0046] Step S701: After the scheduling strategy of the abnormal station is adjusted, the unit period T ’ in which the current time is located is obtained ’ ; ac ) i ;

[0047] Step S702: The expected boarding and alighting frequency (f ex ) i of the i-th station at the current time is obtained ac ) i -(f ex ) i | >= Delta f th , an abnormality alarm is sent to the current scheduling strategy.

[0048] Compared with the prior art, the beneficial effects of the present application are:

[0049] 1. The present application fuses multi-source data into the OD distribution, breaks away from the traditional static model which only relies on the user boarding and alighting data to one-sidedly predict the travel demand, analyzes the influence of various external factors on user travel, effectively reduces the error generated by OD simulation prediction, and improves the accuracy.

[0050] 2. The present application analyzes the deviation of the dimension information of each dimension, compares the boarding and alighting frequency difference in different time periods, analyzes the correlation of various external factors, and stratifies various external factors, can accurately predict based on the actual collected information, and help accurate allocation of vehicle scheduling.

[0051] 3. The present application timely predicts the user travel volume of each station, helps staff to timely mobilize vehicles, at the same time, compares the deviation with the actual travel volume, timely discovers and handles the abnormal situation, avoids the influence of unexpected events on user travel, and effectively improves the user travel efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 It is a step schematic diagram of a travel OD simulation method based on public transportation station boarding volume. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0054] Embodiment: As shown in the figure, the present application provides a travel OD simulation method based on the boarding volume of public transport stations, and the simulation method comprises the following steps: Figure 1

[0055] Step S100: After user authorization, record the stations passed by the user in each ride process, collect multi-dimensional information of each station in different ways during the ride process, generate corresponding ride records, and generate the OD matrix of the user based on all ride records of the user;

[0056] Among them, step S100 comprises the following steps:

[0057] Step S101: Set a boarding volume and an alighting volume for each station, and whenever a user boards public transportation at a certain station, add one to the boarding volume of the certain station, and record each station passed by the user in real time during the ride process until the user alights at another station, and add one to the alighting volume of the another station;

[0058] Step S102: Set a certain station where the user boards public transportation as the initial station and another station where the user alights as the terminal station, sort all stations passed by the user from the initial station to the terminal station in order to generate a ride route of the user, and select a station from the ride route, and obtain the dimensional information of the selected station in several dimensions by using a plurality of preset information collection methods, wherein one information collection method corresponds to the dimensional information of one dimension, and the dimensional information of the selected station in each dimension is obtained;

[0059] Step S103: Summarize the ride route containing each station and the dimensional information of the selected station in each dimension to generate a ride record of the user, and summarize all ride records of the user to generate an OD matrix of the user.

[0060] Step S200: Based on the OD matrix of each user, analyze the distribution of all stations to obtain the boarding and alighting frequency of each station, and select a station at random, analyze the multi-dimensional information recorded in each ride record of the selected station to obtain the association between different dimensional information and the boarding and alighting frequency;

[0061] ​Step S200 includes the following steps:

[0062] Step S201: Obtain the OD matrix of all users, count the number of times each station is the initial station and the terminal station, and set the number of times the i-th station is the initial station to m i and the number of terminal stations is n i ; Count the total number of ride records of all users as Num, and calculate the average boarding and alighting frequency f at the i-th station i =(m i +n i ) / Num;

[0063] Step S202: Obtain the travel time period of each travel record, preset a unit period T, arbitrarily select the j-th unit period, and obtain the time interval of the j-th unit period as TR jT =[(j-1)T,jT); arbitrarily select a travel time period (t1,t2) of a travel record, if (t1,t2)∈TR jT , then the selected riding record is set as the target riding record of the jth unit period;

[0064] Step S203: Extract all target ride records in the jth unit cycle and obtain the boarding and alighting frequency (f i ) j , and obtain the deviation degree of the i-th station (η i ) j =[(f i ) j -f i ] / f i ; Randomly select a target ride record. If the i-th station is the starting station or the terminal station in the selected target ride record, obtain the dimension information of the i-th station in each dimension, extract features from the dimension information of each dimension, and obtain the correlation coefficient g of the extracted features in the j-th unit period j =(η i ) j ;

[0065] Step S204: arbitrarily select a dimension and perform similarity comparison on the features of the selected dimension in each target ride record. If the obtained similarity exceeds the preset similarity threshold, the compared feature is set as a similar feature, and the correlation coefficient of the compared feature is set as a similar correlation coefficient. The average value of the similar correlation coefficient is calculated to obtain the average correlation coefficient (g j ) ave , and matched with the corresponding similar features to obtain the average correlation coefficient between different dimensional information and the frequency of boarding and descending in each dimension.

[0066] Step S300: Based on the correlation of different dimensional information, the influence of any dimensional information is analyzed, and hierarchical judgment conditions are set to perform hierarchical processing on the different dimensional information;

[0067] Wherein, step S300 includes the following steps:

[0068] Step S301: arbitrarily select two unit periods, select dimension information of one dimension from each of the two unit periods, extract similar features between the two unit periods from the same dimension, and calculate the average correlation coefficients corresponding to the similar features to obtain the correlation coefficient difference Δg between the two unit periods;

[0069] Step S302: Obtain the dimensional information corresponding to the same type of features in two unit periods, perform a difference comparison on the dimensional information of the two unit periods, obtain the deviation degree Δp of the same type of features between the two unit periods, and calculate the influence degree y = Δg / Δp of the same type of features;

[0070] Step S303: Preset an expected impact range (y min ,y max ), and get the expected range difference Δy ex =y max -y min , and divide the stratified value range into a levels, then the influence range of the bth level is , if y∈R b , the dimension information of the extracted similar features is divided into the bth level.

[0071] Step S400: assigning corresponding characteristic coefficients to each dimension of information according to the influence and stratification of different dimensional information;

[0072] Step S400 includes the following steps:

[0073] Step S401: arbitrarily select the dimension information of a dimension, obtain each feature under the selected dimension, and obtain the level at which each feature is located. Suppose there are c features in the selected dimension, where the level at which the dth feature is located is the b(d)th level.

[0074] Step S402: Preset an allocable eigenvalue Z, and allocate the eigenvalue Z to each level according to the set ratio, according to the formula:

[0075] ;

[0076] Where b is a positive integer and b∈(1,a); the distributable coefficient of the b(d)th level is calculated as X b(d) ;

[0077] Example 1: The allocable eigenvalue Z is preset to 1, and the dimensional information of each dimension is divided into 4 levels. Then, the allocable coefficient x1 of the first level is obtained as 1×[1 / (1+2+3+4)]=0.1. Similarly, the allocable coefficient x2 of the second level is obtained as 0.2, the allocable coefficient x3 of the third level is obtained as 0.3, and the allocable coefficient x4 of the fourth level is obtained as 0.4.

[0078] Step S403: Set the number of features contained in the b(d)th level to u b(d) , the distribution coefficient of the d-th feature in the selected dimension is X(d)=X b(d) / u b(d) , get the influence degree of the d-th feature as y(d), according to the formula:

[0079] ;

[0080] Calculate the characteristic coefficient V of the dimension information of the selected dimension;

[0081] Example 2: Assume that the selected dimension contains two features, one of which is at the first level and has an influence of 20%, and the other is at the second level and has an influence of 10%; in addition, the first level contains two features and the distributable coefficient is 0.1, and the second level contains two features and the distributable coefficient is 0.2. The characteristic coefficient of the selected dimension is calculated to be V = (20% × 0.1 / 2 + 10% × 0.2 / 2) / 2 = (0.01 + 0.01) / 2 = 0.01.

[0082] Step S500: Constructing a boarding and alighting simulation model to simulate boarding and alighting conditions at each station; training and correcting the boarding and alighting simulation model based on the multi-dimensional information of each ride record;

[0083] Wherein, step S500 includes the following steps:

[0084] Step S501: arbitrarily select a station and a ride record containing the selected station, set the selected ride record as a test record, obtain several ride records whose ride time period is before the test record, and are adjacent and continuous, from all ride records containing the selected station, obtain dimensional information of each dimension in each of the several ride records, and obtain the data change interval of the same feature in any dimension;

[0085] Step S502: Obtain the travel time periods of the plurality of ride records, merge them into a continuous time period, and obtain the boarding and alighting frequency f of the selected station in the continuous time period; obtain the dimension information of each dimension in the test record, perform a deviation comparison with the corresponding data change intervals in the plurality of ride records, and calculate the deviation degree of each dimension;

[0086] Step S503: Set the deviation degree of the kth dimension in the test record as q k , and obtain the characteristic coefficient of the dimension information of the kth dimension as V k , and construct a boarding and alighting simulation model:

[0087] ;

[0088] wherein c ’ is the number of dimensions contained in the test record, s is a constant coefficient; the actual boarding and alighting frequency of the boarding time period of the test record is obtained as f ac , a preset allowed frequency deviation Δf th is obtained, and the numerical range of the constant coefficient s is obtained, such that |f ’ -f ac |≤Δf th ;

[0089] Step S504: Arbitrarily select the remaining boarding records to train the boarding and alighting simulation model, and correct the constant coefficient s. If the corrected constant coefficient s has several values satisfying |f ’ -f ac |≤Δf th , a value is randomly selected to confirm the constant coefficient s.

[0090] In example 3, the average boarding and alighting frequency f of the station in the selected continuous time period is set as 0.5, the dimension information of the station at each dimension at the current time is compared with the dimension information in the continuous time period, and the deviation degrees of two dimensions are obtained as 20% and 30%, respectively, and the characteristic coefficients of the two dimensions are obtained as 0.01 and 0.02, respectively. f ’ =0.5+20%×0.01+30%×0.02+s=0.508+s is calculated, and the boarding and alighting frequency f ’ obtained in the subsequent current time period is 0.55, and s=0.042 is obtained. If the preset allowed frequency deviation is 0.005, the numerical range of s can be (0.037, 0.047), and the numerical range of the constant coefficient s is continuously corrected by the boarding and alighting frequencies of the remaining time periods, and finally the value of the constant coefficient s is determined.

[0091] Step S600: Real-time acquisition of multi-dimensional information of each station, prediction of expected boarding and alighting amount of each station based on the boarding and alighting simulation model, and abnormal identification of each station;

[0092] Step S600 includes the following steps:

[0093] Step S601: Obtain a time interval shortest to the current time, and obtain the landing and taking-off frequency of each station in the several unit periods and the data change interval of each station in each dimension;

[0094] Step S602: Obtain the dimension information of each station in each dimension at the current time, and obtain the deviation degree of each dimension; select an arbitrary station, input the landing and taking-off frequency of the selected station in the several unit periods and the deviation dimension of each dimension into the landing and taking-off simulation model to obtain the expected landing and taking-off frequency f ex of the selected station.

[0095] Step S603: Set the average landing and taking-off frequency of the i-th station as f i , obtain the expected landing and taking-off frequency of the i-th station at the current time as (f ex ) i , and if |f i -(f ex ) i |≥Δf th , mark the i-th station as abnormal and send a dispatching reminder.

[0096] Step S700: Analyze the actual landing and taking-off frequency of each abnormal station after the adjustment of the dispatching strategy, perform abnormal analysis on the adjustment of the dispatching strategy, and perform feedback;

[0097] The step S700 includes the following steps:

[0098] Step S701: After the adjustment of the dispatching strategy for the abnormal station, obtain the unit period T ’ at the current time, and set the actual landing and taking-off frequency of the i-th station in the unit period T ’ as (f ac ) i .

[0099] Step S702: Obtain the expected landing and taking-off frequency of the i-th station at the current time as (f ex ) i , and if |(f ac ) i -(f ex ) i |≥Δf th , send an abnormality reminder for the current dispatching strategy.

[0100] It will be apparent to those skilled in the art that the application is not limited to the details of the above-exemplified embodiments and that the present application can be implemented in other particular forms without departing from the spirit or essential characteristics of the present application. The embodiments should therefore be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the above description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No reference signs in the claims should be considered as limiting the scope of the claims with respect to the figures of the patent document.

Claims

1. A travel OD simulation method based on the number of boardings at public transportation stations, characterized by: The simulation method comprises the following steps: Step S100: After the user's authorization, each station that the user passes through during each ride is recorded. During the ride, multi-dimensional information is collected from each station in different ways to generate corresponding ride records. The user's OD matrix is ​​generated based on all ride records of the user. Step S200: Based on the OD matrix of each user, the distribution of all stations is analyzed to obtain the boarding and alighting frequencies of each station; a station is randomly selected, and the multi-dimensional information recorded in each ride record of the selected station is analyzed to obtain the correlation between different dimensional information and boarding and alighting frequencies; Step S300: Based on the correlation of different dimensional information, the influence of any dimensional information is analyzed, and hierarchical judgment conditions are set to perform hierarchical processing on the different dimensional information; Step S400: assigning corresponding characteristic coefficients to each dimension of information according to the influence and stratification of different dimensional information; Step S500: Constructing a boarding and alighting simulation model to simulate boarding and alighting conditions at each station; training and correcting the boarding and alighting simulation model based on the multi-dimensional information of each ride record; Step S600: acquiring multi-dimensional information of each current station in real time, predicting the expected landing and boarding volume of each station based on the landing and boarding simulation model, and identifying abnormalities at each station; Step S700: Analyze the actual boarding and landing frequencies of each abnormal station after the scheduling strategy is adjusted, perform abnormal analysis on the adjustment of the scheduling strategy, and provide feedback; The step S300 includes the following steps: Step S301: arbitrarily select two unit periods, select dimension information of one dimension from each of the two unit periods, extract similar features between the two unit periods from the same dimension, and calculate the average correlation coefficients corresponding to the similar features to obtain the correlation coefficient difference Δg between the two unit periods; Step S302: Obtain the dimensional information corresponding to the same type of features in two unit periods, perform a difference comparison on the dimensional information of the two unit periods, obtain the deviation degree Δp of the same type of features between the two unit periods, and calculate the influence degree y = Δg / Δp of the same type of features; Step S303: Preset an expected impact range (y min ,y max ), and get the expected range difference Δy ex =y max -y min , and divide the stratified value range into a levels, then the influence range of the bth level is , if y∈R b , then the dimension information of the extracted similar features is divided into the bth level; The step S400 includes the following steps: Step S401: arbitrarily select the dimension information of a dimension, obtain each feature under the selected dimension, and obtain the level at which each feature is located. Suppose there are c features in the selected dimension, where the level at which the dth feature is located is the b(d)th level. Step S402: Preset an allocable eigenvalue Z, and allocate the eigenvalue Z to each level according to the set ratio, according to the formula: ； Where b is a positive integer and b∈(1,a); the distributable coefficient of the b(d)th level is calculated as X b(d) ; Step S403: Set the number of features contained in the b(d)th level to u b(d) , the distribution coefficient of the d-th feature in the selected dimension is X(d)=X b(d) / u b(d) , get the influence degree of the d-th feature as y(d), according to the formula: ； Calculate the characteristic coefficient V of the dimension information of the selected dimension; The step S500 includes the following steps: Step S501: arbitrarily select a station and a ride record containing the selected station, set the selected ride record as a test record, obtain several ride records whose ride time period is before the test record, and are adjacent and continuous, from all ride records containing the selected station, obtain dimensional information of each dimension in each of the several ride records, and obtain the data change interval of the same feature in any dimension; Step S502: Obtain the travel time periods of the plurality of ride records, merge them into a continuous time period, and obtain the boarding and alighting frequency f of the selected station in the continuous time period; obtain the dimension information of each dimension in the test record, perform a deviation comparison with the corresponding data change intervals in the plurality of ride records, and calculate the deviation degree of each dimension; Step S503: Set the deviation degree of the kth dimension in the test record to q k , the characteristic coefficient of the dimension information of the kth dimension is V k , build an ascending and descending simulation model: ； Among them, c ’ is the number of dimensions contained in the test record, s is a constant coefficient; the actual boarding and alighting frequency of the test record during the riding time period is f ac , preset an allowable frequency deviation Δf th , get the numerical range of the constant coefficient s so that |f ’ -f ac |≤Δf th ; Step S504: Randomly select the remaining ride records to train the boarding and alighting simulation model, and modify the constant coefficient s. If the modified constant coefficient s has several values ​​that satisfy |f ’ -f ac |≤Δf th , then randomly select a value to confirm the constant coefficient s.

2. The method for simulating travel OD based on boarding volume at public transportation stations according to claim 1, characterized in that: The step S100 includes the following steps: Step S101: Set a boarding quota and a disembarkation quota for each station. Whenever a user boards public transportation at a station, the boarding quota for the station is incremented by one. Each station the user passes through is recorded in real time during the ride. When the user disembarks at another station, the disembarkation quota for the other station is incremented by one. Step S102: A station where the user boards public transportation is set as the initial station and another station where the user gets off is set as the terminal station, and all stations that the user passes through from the initial station to the terminal station are sorted in order to generate a travel route for the user; a station is randomly selected from the travel route, and dimensional information of the selected station in several dimensions is obtained using several preset information collection methods, wherein one information collection method corresponds to dimensional information of one dimension, and dimensional information of any station in each dimension is obtained; Step S103: Summarize the bus routes including each station and the dimensional information of any station in each dimension to generate a bus record of the user; summarize all the bus records of the user to generate an OD matrix of the user.

3. The travel OD simulation method based on public transportation station boarding volume according to claim 2 is characterized by: The step S200 includes the following steps: Step S201: Obtain the OD matrix of all users, count the number of times each station is the initial station and the terminal station, and set the number of times the i-th station is the initial station to m i and the number of terminal stations is n i ; Count the total number of ride records of all users as Num, and calculate the average boarding and alighting frequency f at the i-th station i =(m i +n i ) / Num; Step S202: Obtain the travel time period of each travel record, preset a unit period T, arbitrarily select the j-th unit period, and obtain the time interval of the j-th unit period as TR jT =[(j-1)T,jT); arbitrarily select a travel time period (t1,t2) of a travel record, if (t1,t2)∈TR jT , then the selected riding record is set as the target riding record of the jth unit period; Step S203: Extract all target ride records in the jth unit cycle and obtain the boarding and alighting frequency (f i ) j , and obtain the deviation degree of the i-th station (η i ) j =[(f i ) j -f i ] / f i ; Randomly select a target ride record. If the i-th station is the starting station or the terminal station in the selected target ride record, obtain the dimension information of the i-th station in each dimension, extract features from the dimension information of each dimension, and obtain the correlation coefficient g of the extracted features in the j-th unit period j =(η i ) j ; Step S204: arbitrarily select a dimension and perform similarity comparison on the features of the selected dimension in each target ride record. If the obtained similarity exceeds the preset similarity threshold, the compared feature is set as a similar feature, and the correlation coefficient of the compared feature is set as a similar correlation coefficient. The average value of the similar correlation coefficient is calculated to obtain the average correlation coefficient (g j ) ave , and matched with the corresponding similar features to obtain the average correlation coefficient between different dimensional information and the frequency of boarding and descending in each dimension.

4. The method for simulating travel OD based on boarding volume at public transportation stations according to claim 3, characterized in that: The step S600 includes the following steps: Step S601: Acquire a number of unit periods with the shortest time interval from the current moment, and obtain the landing and landing frequencies of each station in the unit periods and the data change interval of each station in each dimension; Step S602: Obtain the dimensional information of each station in each dimension at the current moment and obtain the deviation degree of each dimension; arbitrarily select a station, input the landing and landing frequency of the selected station in several unit cycles and the deviation dimension of each dimension into the landing and landing simulation model to obtain the expected landing and landing frequency f of the selected station ex ; Step S603: Set the average boarding and landing frequency of the i-th station to f i , get the expected landing and landing frequency of the i-th station at the current moment (f ex ) i , if |f i -(f ex ) i |≥Δf th , then the i-th site is marked as abnormal and a scheduling reminder is sent.

5. The method for simulating travel OD based on boarding volume at public transportation stations according to claim 4, characterized in that: The step S700 includes the following steps: Step S701: After adjusting the scheduling strategy for the abnormal site, obtain the unit period T of the current time ’ , assuming that the i-th station is in the unit period T ’ The actual landing frequency in (f ac ) i ; Step S702: Obtain the expected landing and landing frequency of the i-th station at the current moment (f ex ) i , if |(f ac ) i -(f ex ) i |≥Δf th , an abnormal reminder will be sent to the current scheduling strategy.

Images