CN110245377A - A kind of travel plan recommended method and recommender system - Google Patents
A kind of travel plan recommended method and recommender system Download PDFInfo
- Publication number
- CN110245377A CN110245377A CN201910378683.7A CN201910378683A CN110245377A CN 110245377 A CN110245377 A CN 110245377A CN 201910378683 A CN201910378683 A CN 201910378683A CN 110245377 A CN110245377 A CN 110245377A
- Authority
- CN
- China
- Prior art keywords
- compartment
- crowding
- time
- travel plan
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 239000011159 matrix material Substances 0.000 claims abstract description 28
- 230000002452 interceptive effect Effects 0.000 claims abstract description 6
- 241001123248 Arma Species 0.000 claims description 18
- 238000001914 filtration Methods 0.000 claims description 18
- 238000005259 measurement Methods 0.000 claims description 9
- 238000010586 diagram Methods 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 8
- 238000012546 transfer Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 5
- 241001269238 Data Species 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000000295 complement effect Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000000739 chaotic effect Effects 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 2
- 230000001174 ascending effect Effects 0.000 claims 1
- 239000000284 extract Substances 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 241000208340 Araliaceae Species 0.000 description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 235000008434 ginseng Nutrition 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/123—Traffic control systems for road vehicles indicating the position of vehicles, e.g. scheduled vehicles; Managing passenger vehicles circulating according to a fixed timetable, e.g. buses, trains, trams
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Radar, Positioning & Navigation (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Remote Sensing (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of travel plan recommended methods, comprising steps of receiving the query information of user；Prediction platform is waited number；Predict compartment crowding；Predict vehicle arrival time；Optimal case judgement: comprehensive platform waiting person number, compartment crowding, arrival time attribute establish standardization decision matrix, assign weight to each attribute, calculate weighting standard decision matrix, the recommendation of optimal travel plan is provided after matching user query information.The invention also discloses a kind of travel plan recommender system, including interactive module, platform are waited number judgment module, crowding judgment module, arrival time judgment module, travel plan determination module.The present invention comprehensively considers the compartment degree of crowding, improves the trip quality of passenger.
Description
Technical field
The present invention relates to technical field of video image processing and mathematical modeling technology field, in particular to are based on digital picture
The travel plan recommended method and recommender system of processing and multiobjective decision-making.
Background technique
With reaching its maturity for the continuous expansion of city size and population, internet and Internet of Things, our daily life
All towards it is intelligent, in terms of diverse development, " smart city " concept becomes the main flow direction of urban construction, and intelligence
Being surging forward for energy traffic system is not only that everyone provides convenience, and also has become current internationally recognized solution traffic problems
Optimal path.Currently, many public transport public platform and APP in real time release one after another, while seizing market, the following wisdom is also indicated that
The development general orientation of public transport must meet common people's demand, solve the problems, such as " pain spot " of existing bus trip.China Shandong,
Zhejiang and other places area starts to try " intelligent bus station " Lai Weicheng by the mass data processing ability of cloud computing and cloud platform
Visitor provides the information service of trip.
In " Intelligent public transportation " demand increasing today, how according to real-time traffic and user query information
It is the major issue that we must think deeply to provide optimal route.In terms of urban public transport intelligent management, how to improve
Passenger satisfaction and improve by bus experience be intelligent Public Transportation System research core content.
The bus trip scheme of the displays such as current public transport APP such as Amap, Baidu map includes line information
(through and transfer), temporal information (arrival time and total time) etc..Chongqing City comprehensive transportation hub (group) Co., Ltd
A kind of transit trip inducible system is disclosed with the patent CN201810858818.5 of Chongqing Jiaotong University, southwestern traffic is big
Patent CN201610193300.5 provides a kind of integration trip integrated decision-making support model, but in terms of experience by bus
Consider not thoughtful enough, the not influence in view of the compartment degree of crowding to experience by bus, such as physical disabilities or band
The passenger of heavy luggage trip be often more likely to wait the low vehicle of a crowding other than riding time.
In addition, the prediction or monitoring for flow of the people are largely all based on macroscopic perspective, i.e., to certain time, centainly
Total flow of the people in range is detected, to be regulated and controled in the angle of macroscopic view.Such as in megastore, railway station, automobile
It stands, the flow of the people monitoring that the scenes such as bustling road carry out.Since its direction range is wider, for ordinary user, specific aim
Not strong, the information that can therefrom extract is extremely limited.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology with it is insufficient, a kind of travel plan recommended method is provided and is pushed away
System is recommended, the method and system comprehensively consider the compartment degree of crowding, improve the trip quality of passenger.
The purpose of the present invention is realized by the following technical solution: a kind of travel plan recommended method, comprising steps of
Receive the query information of user；
Prediction platform is waited number: obtain same route, same platform, same period get on or off the bus number time series,
Prediction platform is waited number；
It predicts compartment crowding: predicting compartment crowding in conjunction with platform waiting person number and compartment demographics；
Predict vehicle arrival time；
Optimal case determines: comprehensive platform waiting person number, compartment crowding, arrival time attribute establish standardization decision
Matrix assigns weight to each attribute, calculates weighting standard decision matrix, provides optimal trip after matching user query information
The recommendation of scheme.This method creatively joined the ginseng that compartment crowding is recommended as scheme in the factor for considering the time
Examine factor, the information content that video obtains in the middle part of propose to get on or off the bus on platform number and compartment is combined to be determined to mark as crowding
Standard, is more accurately obtained the real-time crowding of each bus, so as to more Man's Demands are adapted to, it is more humanized.
Preferably, the prediction platform wait number the step of are as follows:
Number of getting on or off the bus is acquired by the camera of bus front/rear door to extract after data volume reaches a certain level
Same platform, same route, get on the bus number or the number of getting off of same period constitute number sequence；
After examining its stationarity, settling time series model is carried out using the number that gets on and off of the arma modeling to platform
Prediction, while the reference factors such as weather, festivals or holidays are added, keep prediction result more accurate.
Further, under the modeling procedure of the arma modeling:
The stationarity of identifier's Number Sequence: generating timing diagram using the data that get, with the method for observation timing diagram come
The stationarity for determining data, according to the definition of stationarity, stationary sequence has the property of constant mean and constant variance, therefore its
Timing diagram should fluctuate near a constant value, and the range bounded fluctuated；Meanwhile speculating according to the actual situation, the short time
Interior platform number of getting on or off the bus does not have and rises appreciably or decline；
Model identifies and determines rank: the form of ARMA (p, q) model are as follows:
Wherein φ1, φ2..., φp、θ1, θ2..., θqAnd εt, εt-1..., ε1It is auto-regressive parameter, rolling average ginseng respectively
Several and white noise sequence, Xt-1...Xt-pFor the input independent variable sequence of model, that is, the historical basis predicted；According to auto-correlation system
Several and PARCOR coefficients tentatively to judge the order p and q of arma modeling, calculate sample autocorrelation coefficient (ACF) He Pianxiang
Relationship number (PACF)；
The parameter Estimation of model: according to the model and its order of identification, parameter is carried out to model using least square method
Estimation；
Model testing: by model identification, determining rank and parameter Estimation, needs to carry out residual sequence check analysis to model；
When residual sequence is white noise sequence, model of fit is just effective, can be used to predict；Calculate the s step of residual sequence { ε (t) } certainly
Correlation coefficient ρ1、ρ2、...ρs, then construct chi-square statistics amount Fs, it is shown below, FsObey the Χ that freedom degree is s2Distribution；Formula
In, n is the capacity of residual sequence；
It is predicted using arma modeling calculated result；After the data such as incoming line, platform, time, prediction can be obtained
Number of waiting；Using arma modeling calculated result as main basis for forecasting, at the same be added weather, festivals or holidays etc. refer to because
Element keeps prediction result more accurate.
Preferably, the step of prediction compartment crowding are as follows:
By camera in the middle part of compartment, picture portion HOG feature, the pre- crowding for determining compartment are extracted；
By car door camera, the number of people is captured, counts number of getting on or off the bus, by the way that each platform number of getting on the bus adds up, together
When subtract the number of getting off method, obtain number in compartment；
Result and compartment number are determined in conjunction with pre-, calculate crowding of arriving at a station.
The two kinds of crowding analyses of demographics and image analysis complement one another, and the accuracy of crowding judgement can be improved.
Further, the extraction picture portion HOG feature, the pre- crowding for determining compartment specifically:
Using histograms of oriented gradients (Histogram of Oriented Gradient, HOG) feature extraction, information content
It calculates, to video image in the middle part of the compartment of input, its HOG feature is extracted to each picture portion, certain threshold value is set and filters out
Effective gradient value divides subregion to image to count the angular histogram of effective gradient, and each subregion takes g angle direction,
And calculate the gradient value of this g angle；In conjunction with the variance G of effective gradient value2It is calculated with image information entropy H as follows:
H indicates h-th of angle direction in g angle direction in above formula；μ is average gradient value；PijIndicate certain pixel position
The comprehensive characteristics of the gray value and its surrounding pixel intensity profile set；Select the field gray average of image as intensity profile
Space eigenvalues, the pixel grey scale composition characteristic binary group with image is denoted as (i, j), and wherein i indicates the gray value of pixel, 0
≤ i≤255, j indicate field gray value, 0≤j≤255；F (i, j) is characterized the frequency of binary group (i, j) appearance；U, V distinguishes
For the length and width of image；The comentropy of image is H；
The confusion degree E=(G, H) for calculating final image pixel, when subregion gradient value variance is small and comentropy is big, table
The pixel Distribution value of the bright subregion is more chaotic, then the subregion degree of crowding is big；Platform is waited number and compartment demographics
As a result, be complementary to one another with the judgement result of crowding in the middle part of compartment, when crowded, corresponding demographics value also answer it is larger, therefore,
Can demographics and middle part camera crowding calculated result, final compartment crowding is determined to cooperate with, as travel plan
Judgment basis.
Further, the step of prediction compartment crowding specifically:
It calculates real-time crowding: mainly being determined according to the accumulated value of the number of getting on or off the bus after dispatching a car, while utilizing compartment
The crowding that middle part camera calculates carrys out correction result, wherein by camera in the middle part of compartment, extracts picture portion HOG feature
Come assist it is pre- determine compartment crowding, calculate it is as follows:
Wherein, Ereal_timeIt is the confusion degree of real-time query image pixel, i.e., real-time crowding；It is hair
The accumulated value of number of getting on or off the bus after vehicle；Area is the area that can stand；EDIPIt is to be mixed by the image that camera in the middle part of compartment determines
The result of random degree；
Calculating is arrived at a station crowding: by real-time crowding Ereal_timeWhere cumulative upper enquiring vehicle between website and passenger's website
Platform, this intermediate stations gets on or off the bus number historical statistics result to predict to obtain, can be realized to the pre- of crowding of arriving at a station
It surveys, calculates as follows:
Wherein, Egetin_stationIt is crowding of arriving at a station,It is the website that inquires between website where passenger
Number of getting on or off the bus.
Preferably, the step of prediction vehicle arrival time are as follows:
Using the matched method of historic state, under the conditions of originating data is matched, current data and historical data are calculated
Euclidean distance, to predict arrival time；
Distribution-free regression procedure is found and the most similar collection of current input state by the search to a large amount of historical datas
It closes, the state of subsequent time is predicted with this；Assuming that bus reaches the station k-1, originating moment corresponding data mode is S；
Use TkmAnd T'kmIt respectively indicates the m articles that all data modes in historical data base are S and records the station that the corresponding station k-1 k stands
Between running time and k station berthing time；Taking J is neighbour's number of non parametric regression, and D is the dimension of non parametric regression；Using Europe
Family name's distance amIt is matched with historical data, J neighbour is chosen, indicates the set of this J neighbour with A here, it is close according to these
Adjoint point predicts the running time T that bus is stood from the station k k+1 with average weighted methodk:
E is the Neighbor Points in the set A different from m；Similarly, k-d ... can be passed through, when the stop at the station k-2, k-1
Between, predict that the berthing time that the bus is stood in k, d are the website number differed with site k using above-mentioned formula.
Further, predict vehicle arrival time when, by bus when set out in advance to make arrangements prediction berthing time with arrive it is next
The prediction running time stood and be prediction runing time, as the input parameter of Kalman filtering algorithm, to realize to pre-
Survey the dynamic corrections of result；
Bus -1 station of kth prediction berthing time and -1 station of kth to k stand prediction running time and be predict
Runing time, the input parameter as Kalman filtering algorithm；
Kalman filtering algorithm is that the state equation and observational equation by establishing make satisfaction most to signal to be treated
The estimation of small mean square error estimates state variable in conjunction with estimated value and the observation at the current time completion of last moment
Meter, and solve current estimated value；The state equation of the Kalman filtering of bus arrival time prediction is as follows:
Wherein t(k+1)、t'(k+1)Respectively indicate bus from website reach the kth website time experienced and observation when
Between, u (k) indicates that running time between the station that bus is stood from k to k+1, S (k), I (k) and M (k) they are respectively that data mode transfer becomes
Amount, input variable, measurement value coefficient, w (k) and v (k) are respectively input noise and measurement noise, and set Kalman filtering
The coefficient of state transfering variable, input variable and measured value is all 1, and input noise and measurement noise are irrelevant, mean values
For 0 independent white noise；
When bus reaches k-th of website, Kalman filtering is according to the observation T of k-th of websitek m+1With preceding k-1
Basal latency sequence calculates the optimal estimation value of k-th of website, and then obtains the adjusted value of+1 website of kth, according to adjusted value
The time of subsequent website is successively updated, and this updated time series is added in basal latency sequence.
Preferably, the step of optimal case determines are as follows:
Standardize decision matrix；Setting out row program decisions matrix is Z=(zxy) LW, wherein x is travel plan subscript, y
Scheme attribute subscript, zxyIndicate the value of x scheme y attribute, L, W respectively represent the length and width of decision matrix, i.e., the sum of scheme with
Each scheme attribute sum；Scheme attribute includes the contents such as compartment crowding, number of waiting, arrival time, number of transfer.If specification
Change decision matrix B=(bxy) LW, wherein bxyThe value of x scheme y attribute after indicating standardization, then have:
X=1,2..., L；Y=1,2..., W；
Calculate weighted normal decision matrix；Weight is set according to the significance level for influencing trip factors of experience, wherein is arrived at a station
The weight of the maximum weight of time, number of waiting is minimum, constitutes weight vectors Q=[r1,r2,...,rW]T, and then constitute weighting
Normal matrix C=(cxy) LW, wherein cxy=rx·bxy(x=1,2..., L；Y=1,2..., W；R indicates weight coefficient)；
Determine ideal solution and minus ideal result；Determine positive ideal solution cy +(y-th of attribute of positive ideal solution) and minus ideal result cy -
(y-th of attribute of minus ideal result) then:
Positive ideal solutionY=1,2..., W；
Minus ideal resultY=1,2..., W；
Wherein, compartment crowding, number of waiting, arrival time are cost-effectivenes attribute, and positive ideal solution is all minimized, and are born
Ideal solution is all maximized；
Calculate each travel plan to ideal solution and minus ideal result distance；
Distance to positive ideal solution isX=1,2...L；
Distance to minus ideal result isX=1,2...L；
Calculate the queuing index value of each scheme, i.e. comprehensive evaluation index: by being total to the distance of positive ideal solution and minus ideal result
It is same to provide, the comprehensive evaluation index queuing value Que of schemexCalculation method it is as follows:
X=1,2..., L；
By QuexThe superiority and inferiority order of descending arrangement scheme: by the comprehensive evaluation index value of each scheme finally obtained,
That is QuexAscending arrangement, wherein QuexMaximum scheme is optimal case, i.e. the final output of this algorithm.
Further, the travel plan of the recommendation is from being most matched to a series of general suggested designs of matching degree.
A kind of travel plan recommender system, including wait number judgment module, crowding of interactive module, platform judge mould
Block, arrival time judgment module, travel plan determination module；
The interactive module is used to receive the query information of user, pushes optimal travel plan to user；
The platform is waited number judgment module, and for obtaining same route, same platform, same period get on or off the bus people
Number time serieses, prediction platform are waited number；
The crowding judgment module, for predicting compartment crowding in conjunction with platform waiting person number and compartment demographics；
The arrival time judgment module, for predicting vehicle arrival time；
The travel plan determination module is built for integrating platform waiting person number, compartment crowding, arrival time attribute
Vertical standardization decision matrix, assigns weight to each attribute, calculates weighting standard decision matrix, after matching user query information
Provide the recommendation of optimal travel plan.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1, the present invention creatively joined the reference that compartment crowding is recommended as scheme in the factor for considering the time
Factor, the information content that video obtains in the middle part of propose to get on or off the bus on platform number and compartment is combined to be determined to mark as crowding
Standard, is more accurately obtained the real-time crowding of each bus, so as to more Man's Demands are adapted to, it is more humanized.
2, the present invention prediction platform wait number when settling time series model, using arma modeling to the upper and lower of platform
Vehicle number is predicted, while the reference factors such as weather, festivals or holidays are added, and keeps prediction result more accurate.
3, when present invention prediction compartment crowding, judge that compartment crowding, the two are mutual in conjunction with the number of getting on or off the bus and HOG image
For supplement, the accuracy of crowding judgement can be improved.
4, the travel plan recommendation that the present invention provides is that a series of go on a journey from excellent to general is recommended, and fully takes into account passenger
Demand while ensure that the possibility of its unrestricted choice.
Detailed description of the invention
Fig. 1 is travel plan recommender system schematic diagram of the present invention.
Fig. 2 is that crowding judgment module of the present invention judges process.
Fig. 3 is that travel plan recommended method of the present invention predicts the time flow chart that arrives at a station.
Fig. 4 is travel plan recommended method flow chart of the present invention.
Specific embodiment
For a better understanding of the technical solution of the present invention, the implementation that the present invention is described in detail provides with reference to the accompanying drawing
Example, embodiments of the present invention are not limited thereto.
Embodiment
When user selects the destination of trip, system is accustomed to according to the trip of user, as certain customers are more biased towards trip
Comfort level, that is, wish that there is a lower crowding in compartment, the weight of each attribute of real-time update assigns higher for compartment crowding
Weight, after prediction crowding and the arrival time by obtaining each scheme compartment, using TOPSIS method to all schemes into
Row sequence, recommends user for optimal case.
Present embodiment discloses a kind of bus trip proposal recommending method, all steps of the present embodiment are all based on JAVA
Develop what environment, OpenCV function library and MySQL database were completed.Specific implementation technical solution of the invention is that input is used first
For the query information at family to the interactive module of travel plan recommender system, travel plan recommender system further includes that the platform number of waiting is sentenced
Disconnected module, crowding judgment module, arrival time judgment module, travel plan determination module, when user's input riding information, i.e.,
It is waited number, compartment crowding and arrival time according to platform, come that COMPREHENSIVE CALCULATING goes out optimal travel plan is by bus in real time
System.
(1) crowding judge, mainly using histograms of oriented gradients (Histogram of Oriented Gradient,
HOG) feature extraction, information computing, and compartment crowding is determined in conjunction with platform waiting person number and compartment demographics.To defeated
Video image in the middle part of the compartment entered, extracts its HOG feature to each picture portion, sets certain threshold value and filter out effective gradient
Value divides subregion to image, each subregion takes g angle direction, and calculates this to count the angular histogram of effective gradient
The gradient value of g angle.In conjunction with the variance G of effective gradient value2It is calculated with image information entropy H as follows:
H indicates h-th of angle direction in g angle direction；μ is average gradient value, above formula PijIndicate certain location of pixels
On gray value and its surrounding pixel intensity profile comprehensive characteristics, select the field gray average of image as intensity profile
Space eigenvalues, the pixel grey scale composition characteristic binary group with image, are denoted as (i, j), wherein i indicate pixel gray value (0≤
I≤255), j indicates field gray value (0≤j≤255), and f (i, j) is characterized the frequency of binary group (i, j) appearance, U, V difference
For the length and width of image, then the comentropy of image is H.
The confusion degree that final image pixel is calculated by E=(G, H), when subregion gradient value variance is small and comentropy
Greatly, all show that the pixel Distribution value of the subregion is more chaotic, then the subregion degree of crowding is big.Platform is waited number and compartment number
Statistics as a result, be complementary to one another with the judgement result of crowding in the middle part of compartment, when crowded, corresponding demographics value also should be compared with
Greatly, therefore, the demographics of compartment front/rear door camera and middle part camera crowding calculated result determine finally to cooperate with
Compartment crowding, the judgment basis as travel plan.
Specifically, the step of prediction compartment crowding are as follows:
It calculates real-time crowding: mainly being determined according to the accumulated value of the number of getting on or off the bus after dispatching a car, while utilizing compartment
The crowding that middle part camera calculates carrys out correction result, wherein by camera in the middle part of compartment, extracts picture portion HOG feature
Come assist it is pre- determine compartment crowding, calculate it is as follows:
Wherein, Ereal_timeIt is the confusion degree of real-time query image pixel, i.e., real-time crowding；It is hair
The accumulated value of number of getting on or off the bus after vehicle；Area is the area that can stand；EDIPIt is to be mixed by the image that camera in the middle part of compartment determines
Random degree result；
Calculating is arrived at a station crowding: by real-time crowding Ereal_timeWhere cumulative upper enquiring vehicle between website and passenger's website
Platform, this intermediate stations gets on or off the bus number historical statistics result to predict to obtain, can be realized to the pre- of crowding of arriving at a station
It surveys, calculates as follows:
Wherein, Egetin_stationIt is crowding of arriving at a station,It is the website that inquires between website where passenger
Number of getting on or off the bus.
(2) in bus arrival time predicted portions, the present embodiment is based on a large amount of bus arrival historical data, passes through
Distribution-free regression procedure obtains arrival time predicted value, and the input parameter as Kalman filtering algorithm is to realize to arrival time
Dynamic corrections with the public transport arrival time that more calculates to a nicety.
Distribution-free regression procedure is found and the most similar collection of current input state by the search to a large amount of historical datas
It closes, the state of subsequent time is predicted with this.Assuming that bus reaches the station k-1, originating moment corresponding data mode is S.
Use TkmAnd T'kmIt respectively indicates the m articles that all data modes in historical data base are S and records the station that the corresponding station k-1 k stands
Between running time and k station berthing time.Taking J is neighbour's number of non parametric regression, and D is the dimension of non parametric regression.Using Europe
Family name's distance amIt is matched with historical data, chooses J neighbour's (indicating the set of this J neighbour with A here), it is close according to these
Adjoint point predicts the running time T that bus is stood from the station k k+1 with average weighted methodk:
Similarly, k-d ... can be passed through, the berthing time at the station k-2, k-1 predicts that the bus exists using above-mentioned formula
The berthing time at the station k, d is the website number differed with site k.Prediction berthing time and kth -1 station of the bus at -1 station of kth
The prediction running time stood to k and to predict runing time, as Kalman filtering algorithm input parameter.
Kalman filtering algorithm is that the state equation and observational equation by establishing make satisfaction most to signal to be treated
The estimation of small mean square error estimates state variable in conjunction with estimated value and the observation at the current time completion of last moment
Meter, and solve current estimated value.The state equation of the Kalman filtering of bus arrival time prediction is as follows:
Wherein t(k+1)、t'(k+1)Respectively indicate bus from website reach the kth website time experienced and observation when
Between, u (k) indicates that running time between the station that bus is stood from k to k+1, S (k), I (k) and M (k) they are respectively that data mode transfer becomes
Amount, input variable, measurement value coefficient, w (k) and v (k) are respectively input noise and measurement noise, and set Kalman filtering
The coefficient of state transfering variable, input variable and measured value is all 1, and input noise and measurement noise are irrelevant, mean values
For 0 independent white noise.
When bus reaches k-th of website, Kalman filtering is according to the observation T of k-th of websitek m+1With preceding k-1
Basal latency sequence calculates the optimal estimation value of k-th of website, and then obtains the adjusted value of+1 website of kth, according to adjusted value
The time of subsequent website is successively updated, and this updated time series is added in basal latency sequence.
From the point of view of the historical data acquired at present according to us, running time extremely has rule between the station of the bus of same route
Rule can be followed, therefore under the premise of having a large amount of historical datas, we can find with the most similar set of current input state, with
This predicts the state of subsequent time, and then predicts arrival time, and kalman filter method be utilized it is the most neighbouring
Historical data is modified, and has preferable real-time, avoids use data bring relatively remote pre- to a certain extent
Error is surveyed, the two method complements each other, so as to the head office that can predict more accurate arrival time and passenger arrives at the destination
Sail the time.
(3) number of waiting prediction, the data that we will acquire are drawn according to route, platform, number of weeks, period
Point, obtaining the same route same platform same period gets on or off the bus number time series.Establish arma modeling, and then reciprocity vehicle people
Number is predicted.Under modeling procedure:
1. the stationarity of checking sequence: generating timing diagram using the data got, sentenced with the method for observation timing diagram
The stationarity of fixed number evidence, according to the definition of stationarity, stationary sequence has the property of constant mean and constant variance, therefore at that time
Sequence figure should fluctuate near a constant value, and the range bounded fluctuated.Meanwhile speculating according to the actual situation, in the short time
Platform number of getting on or off the bus does not have and rises appreciably or decline.
2. model identifies and determines rank: the form of general ARMA (p, q) model are as follows:
Whereinθ1, θ2..., θqAnd εt, εt-1..., ε1It is auto-regressive parameter, rolling average respectively
Parameter and white noise sequence, Xt-1...Xt-pFor the input independent variable sequence of model, that is, the historical basis predicted.According to auto-correlation
Coefficient and PARCOR coefficients tentatively to judge the order p and q of arma modeling, calculate sample autocorrelation coefficient (ACF) and inclined
Related coefficient (PACF).
3. the parameter Estimation of model: according to the model and its order of identification, being joined using least square method to model
Number estimation.
4. model testing: by model identification, determining rank and parameter Estimation, need to carry out residual sequence to model to examine to divide
Analysis.When residual sequence is white noise sequence, model of fit is just effective, can be used to predict.Calculate the s step of difference sequence { ε (t) }
Auto-correlation coefficient ρ1、ρ2、...ρs, then construct chi-square statistics amount Fs, it is shown below, FsObey the Χ that freedom degree is s2Distribution；
In formula, n is the capacity of residual sequence；
5. being predicted using arma modeling calculated result.After the data such as incoming line, platform, time, it can be obtained pre-
The number of waiting surveyed.Using arma modeling calculated result as main basis for forecasting, at the same be added weather, festivals or holidays etc. refer to because
Element keeps prediction result more accurate.
(4) travel plan determines, by comprehensively considering compartment crowding, when platform waits number with vehicle and public transport is arrived at a station
Between this three Xiang Zhibiao, come scientifically and rationally for passenger formulate riding scheme, improve ride satisfaction, realize intelligent travel.
TOPSIS is the sort method of similarity to ideal solution, it gives by the ideal solution and minus ideal result of decision-making problem of multi-objective
Each schemes ranking in scheme set.By to the compartment crowding of each public transport, number of waiting, these three indexs of arrival time into
Row quantization, and respective weights are assigned according to the actual situation, each public transport is calculated separately out to ideal solution at a distance from minus ideal result,
By each schemes ranking, optimal solution is provided.Specific step is as follows:
1. standardizing decision matrix；Setting out row program decisions matrix is Z=(zxy) LW, wherein x is travel plan subscript,
Y is scheme attribute subscript, zxyIndicate the value of x scheme y attribute, L, W respectively represent the length and width of decision matrix, the i.e. sum of scheme
It is total with each scheme attribute, similarly hereinafter.Scheme attribute includes in compartment crowding, number of waiting, arrival time, number of transfer etc.
Hold.If specified decision matrix B=(bxy) LW, wherein bxyThe value of x scheme y attribute after indicating standardization, then have:
X=1,2..., L；Y=1,2..., W；
2. calculating weighted normal decision matrix；Weight is set according to the significance level for influencing trip factors of experience, wherein is arrived
It stands the maximum weight of time, the weight for number of waiting is minimum, constitutes weight vectors Q=[r1,r2,...,rW]T, and then constitute and add
Weigh normal matrix C=(cxy) LW, wherein cxy=rx·bxy, x=1,2..., L；Y=1,2..., W；R indicates weight coefficient；
3. determining ideal solution and minus ideal result；Determine positive ideal solution cy +(y-th of attribute of positive ideal solution) and minus ideal result
cy -(y-th of attribute of minus ideal result) then:
Positive ideal solutionY=1,2..., W；
Minus ideal resultY=1,2..., W；
Wherein, compartment crowding, number of waiting, arrival time are cost-effectivenes attribute, and positive ideal solution is all minimized, and are born
Ideal solution is all maximized.
4. calculate each travel plan to ideal solution and minus ideal result distance；
Distance to positive ideal solution isX=1,2...L；
Distance to minus ideal result isX=1,2...L；
5. calculating the queuing index value (i.e. comprehensive evaluation index) of each scheme: by the distance to positive ideal solution and minus ideal result
It provides jointly, calculation method is as follows:
X=1,2..., L；
6. pressing QuexThe superiority and inferiority order of descending arrangement scheme: by the comprehensive evaluation index of each scheme finally obtained
Value, i.e. QuexAscending arrangement, wherein QuexMaximum scheme is optimal case, i.e. the final output of this algorithm.
Since the transfer of public transport can also be applied on subway with analogy to subway, the present embodiment method.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (10)
1. a kind of travel plan recommended method, which is characterized in that comprising steps of
Receive the query information of user；
Prediction platform is waited number: obtaining same route, same platform, same period get on or off the bus number time series, prediction
Platform is waited number；
It predicts compartment crowding: predicting compartment crowding in conjunction with platform waiting person number and compartment demographics；
Predict vehicle arrival time；
Optimal case determines: comprehensive platform waiting person number, compartment crowding, arrival time attribute establish standardization decision matrix,
Weight is assigned to each attribute, calculates weighting standard decision matrix, provides optimal travel plan after matching user query information
Recommendation.
2. travel plan recommended method according to claim 1, which is characterized in that the prediction platform is waited the step of number
Suddenly are as follows:
Number of getting on or off the bus is acquired by the camera of bus front/rear door to extract same after data volume reaches a certain level
Platform, same route, get on the bus number or the number of getting off of same period constitute number sequence；
After examining its stationarity, settling time series model is carried out pre- using the number that gets on and off of the arma modeling to platform
It surveys, while weather, festivals or holidays reference factor is added.
3. travel plan recommended method according to claim 2, which is characterized in that the modeling procedure of the arma modeling
Under:
The stationarity of identifier's Number Sequence: generating timing diagram using the data got, is determined with the method for observation timing diagram
The stationarity of data, according to the definition of stationarity, stationary sequence has the property of constant mean and constant variance, therefore its timing
Figure should fluctuate near a constant value, and the range bounded fluctuated；Meanwhile speculating according to the actual situation, station in the short time
Platform number of getting on or off the bus does not have and rises appreciably or decline；
Model identifies and determines rank: the form of ARMA (p, q) model are as follows:
Wherein φ1, φ2..., φp、θ1, θ2..., θqAnd εt, εt-1..., ε1Be respectively auto-regressive parameter, rolling average parameter and
White noise sequence, Xt-1...Xt-pFor the input independent variable sequence of model, that is, the historical basis predicted；According to auto-correlation coefficient and
PARCOR coefficients tentatively to judge the order p and q of arma modeling, calculate sample autocorrelation coefficient and partial correlation coefficient；
The parameter Estimation of model: according to the model and its order of identification, parameter Estimation is carried out to model using least square method；
Model testing: by model identification, determining rank and parameter Estimation, needs to carry out residual sequence check analysis to model；When residual
Model of fit is just effective when difference sequence is white noise sequence, can be used to predict；The s for calculating residual sequence { ε (t) } walks auto-correlation
Coefficient ρ1、ρ2、...ρs, then construct chi-square statistics amount Fs, it is shown below, FsObey the X that freedom degree is s2Distribution；In formula, n
For the capacity of residual sequence；
It is predicted using arma modeling calculated result；After the data such as incoming line, platform, time, the time of prediction can be obtained
Vehicle number；Using arma modeling calculated result as main basis for forecasting, while weather, festivals or holidays reference factor is added.
4. travel plan recommended method according to claim 1, which is characterized in that the step of the prediction compartment crowding
Are as follows:
By camera in the middle part of compartment, picture portion HOG feature, the pre- crowding for determining compartment are extracted；
By car door camera, the number of people is captured, number of getting on or off the bus is counted and is subtracted simultaneously by the way that each platform number of getting on the bus adds up
The method for going the number of getting off obtains number in compartment；
Result and compartment number are determined in conjunction with pre-, calculate crowding of arriving at a station.
5. travel plan recommended method according to claim 4, which is characterized in that the extraction picture portion HOG feature,
The pre- crowding for determining compartment specifically:
Using histograms of oriented gradients, feature extraction, information computing, to video image in the middle part of the compartment of input, to each figure
As its HOG feature of multi-subarea extracting, sets certain threshold value and filter out effective gradient value, to count the angle histogram of effective gradient
Figure divides subregion to image, and each subregion takes g angle direction, and calculates the gradient value of this g angle；In conjunction with effective gradient
The variance G of value2It is calculated with image information entropy H as follows:
H indicates h-th of angle direction in g angle direction in above formula；μ is average gradient value；PijIt indicates on certain location of pixels
Gray value and its surrounding pixel intensity profile comprehensive characteristics；Select sky of the field gray average of image as intensity profile
Between characteristic value, the pixel grey scale composition characteristic binary group with image is denoted as (i, j), and wherein i indicates the gray value of pixel, 0≤i
≤ 255, j expression field gray value, 0≤j≤255,；F (i, j) is characterized the frequency of binary group (i, j) appearance；U, V is respectively
The length and width of image；The comentropy of image is H；
The confusion degree E=(G, H) for calculating final image pixel shows this when subregion gradient value variance is small and comentropy is big
The pixel Distribution value of subregion is more chaotic, then the subregion degree of crowding is big；Platform wait number and compartment demographics as a result,
Be complementary to one another with the judgement result of crowding in the middle part of compartment, when crowded, corresponding demographics value also answer it is larger, therefore, compartment
The demographics and middle part camera crowding calculated result of front/rear door camera determine final compartment crowding to cooperate with,
Judgment basis as travel plan.
6. travel plan recommended method according to claim 1, which is characterized in that the step of the prediction vehicle arrival time
Suddenly are as follows:
Using the matched method of historic state, under the conditions of originating data is matched, the Europe of current data and historical data is calculated
Family name's distance, to predict arrival time；
Distribution-free regression procedure by the search to a large amount of historical datas, find with the most similar set of current input state,
The state of subsequent time is predicted with this；Assuming that bus reaches the station k-1, originating moment corresponding data mode is S；With
TkmAnd T'kmThe m articles for respectively indicating that all data modes in historical data base are S record corresponding k-1 stand the station k station between
The berthing time of running time and the station k；Taking J is neighbour's number of non parametric regression, and D is the dimension of non parametric regression；Using Euclidean
Distance amIt is matched with historical data, J neighbour is chosen, the set of this J neighbour is indicated with A here, according to these neighbours
Point predicts the running time T that bus is stood from the station k k+1 with average weighted methodk:
E is the Neighbor Points in the set A different from m；
Similarly, k-d ... can be passed through, the berthing time at the station k-2, k-1 predicts that the bus is stood in k using above-mentioned formula
Berthing time, d is the website number differed with site k.
7. travel plan recommended method according to claim 6, which is characterized in that, will be public when prediction vehicle arrival time
Hand over vehicle when set out in advance to make arrangements prediction berthing time with to the next stop prediction running time and for prediction runing time, as karr
The input parameter of graceful filtering algorithm, to realize the dynamic corrections to prediction result；
Bus -1 station of kth prediction berthing time and -1 station of kth to k stand prediction running time and for predict run
Time, the input parameter as Kalman filtering algorithm；
Kalman filtering algorithm is that the state equation and observational equation by establishing make satisfaction minimum to signal to be treated
The estimation of square error completes the estimation to state variable in conjunction with the estimated value and the observation at current time of last moment, and
Solve current estimated value；The state equation of the Kalman filtering of bus arrival time prediction is as follows:
Wherein t(k+1)、t'(k+1)It respectively indicates bus website from and reaches kth website time experienced and observation time, u
(k) indicate running time between the station stood from k to k+1 of bus, S (k), I (k) and M (k) be respectively data mode transfering variable,
Input variable, measurement value coefficient, w (k) and v (k) are respectively input noise and measurement noise, and set the state of Kalman filtering
The coefficient of transfering variable, input variable and measured value is all 1, and it is 0 that input noise and measurement noise, which are irrelevant, mean values,
Independent white noise；
When bus reaches k-th of website, Kalman filtering is according to the observation T of k-th of websitek m+1When with preceding k-1 basis
Between sequence calculate the optimal estimation value of k-th of website, and then obtain the adjusted value of+1 website of kth, successively more according to adjusted value
The time of new subsequent website, and this updated time series is added in basal latency sequence.
8. travel plan recommended method according to claim 1, which is characterized in that the step of optimal case determines
Are as follows:
Standardize decision matrix；Setting out row program decisions matrix is Z=(zxy)L·W, wherein x is travel plan subscript, and y is scheme
Attribute subscript, zxyIndicate the value of x scheme y attribute, L, W respectively represent the length and width of decision matrix, the i.e. sum of scheme and each side
Case attribute sum；Scheme attribute includes compartment crowding, number of waiting, arrival time, number of transfer；If specified decision matrix
B=(bxy)L·W, wherein bxyThe value of x scheme y attribute after indicating standardization, then have:
Calculate weighted normal decision matrix；Weight is set according to the significance level for influencing trip factors of experience, wherein arrival time
Maximum weight, the weight of number of waiting is minimum, constitutes weight vectors Q=[r1,r2,...,rW]T, and then constitute weighted normal
Matrix C=(cxy)L·W, wherein cxy=rx·bxy, x=1,2..., L；Y=1,2..., W；R indicates weight coefficient；
Determine ideal solution and minus ideal result；Determine positive ideal solution cy +With minus ideal result cy -；Then:
Positive ideal solution
Minus ideal result
Wherein, compartment crowding, number of waiting, arrival time are cost-effectivenes attribute, and positive ideal solution is all minimized, and bear ideal
Solution is all maximized；
Calculate each travel plan to ideal solution and minus ideal result distance；
Distance to positive ideal solution is
Distance to minus ideal result is
It calculates the queuing index value of each scheme, i.e. comprehensive evaluation index: being given jointly by the distance to positive ideal solution and minus ideal result
Out, the comprehensive evaluation index queuing value Que of schemexCalculation method it is as follows:
By QuexThe superiority and inferiority order of descending arrangement scheme: by the comprehensive evaluation index value of each scheme finally obtained, i.e. Quex
Ascending arrangement, wherein QuexMaximum scheme is optimal case, i.e. the final output of this algorithm.
9. travel plan recommended method according to claim 8, which is characterized in that the travel plan of the recommendation is from most
It is matched to a series of general suggested designs of matching degree.
10. a kind of travel plan recommender system, which is characterized in that wait number judgment module, crowded including interactive module, platform
Spend judgment module, arrival time judgment module, travel plan determination module；
The interactive module is used to receive the query information of user, pushes optimal travel plan to user；
The platform is waited number judgment module, for obtaining same route, same platform, same period get on or off the bus number when
Between sequence, prediction platform waits number；
The crowding judgment module, for predicting compartment crowding in conjunction with platform waiting person number and compartment demographics；
The arrival time judgment module, for predicting vehicle arrival time；
The travel plan determination module establishes mark for integrating platform waiting person number, compartment crowding, arrival time attribute
Standardization decision matrix assigns weight to each attribute, calculates weighting standard decision matrix, provides after matching user query information
The recommendation of optimal travel plan.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910378683.7A CN110245377B (en) | 2019-05-08 | 2019-05-08 | Travel scheme recommendation method and recommendation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910378683.7A CN110245377B (en) | 2019-05-08 | 2019-05-08 | Travel scheme recommendation method and recommendation system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110245377A true CN110245377A (en) | 2019-09-17 |
| CN110245377B CN110245377B (en) | 2023-08-04 |
Family
ID=67883816
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910378683.7A Active CN110245377B (en) | 2019-05-08 | 2019-05-08 | Travel scheme recommendation method and recommendation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110245377B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782694A (en) * | 2019-10-31 | 2020-02-11 | 同济大学 | Internet bus arrival and stop management optimization method |
| CN111091231A (en) * | 2019-11-25 | 2020-05-01 | 珠海格力电器股份有限公司 | Prediction model training method, time prediction method, training device and terminal |
| CN111415679A (en) * | 2020-03-25 | 2020-07-14 | Oppo广东移动通信有限公司 | Site identification method, device, terminal and storage medium |
| CN112256967A (en) * | 2020-10-27 | 2021-01-22 | 北京字跳网络技术有限公司 | Travel recommendation method, device, equipment and storage medium |
| CN114299746A (en) * | 2021-12-30 | 2022-04-08 | 武汉长飞智慧网络技术有限公司 | Image recognition-based bus scheduling method, device, equipment and medium |
| CN115588298A (en) * | 2022-10-28 | 2023-01-10 | 广州地铁集团有限公司 | Urban rail passenger flow broadcasting induction method based on machine vision |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140012498A1 (en) * | 2012-03-19 | 2014-01-09 | Thomas F. Gustafson | Informed Traveler Program and Application |
| JP2014010493A (en) * | 2012-06-27 | 2014-01-20 | Koito Electric Industries Ltd | Bus stop information service system |
| CN105096584A (en) * | 2014-05-06 | 2015-11-25 | 中兴通讯股份有限公司 | Traffic decision support method, device, and system |
| CN105702035A (en) * | 2016-03-23 | 2016-06-22 | 武汉大学 | Method for evaluating bus taking difficulty degree by using historical bus data |
-
2019
- 2019-05-08 CN CN201910378683.7A patent/CN110245377B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140012498A1 (en) * | 2012-03-19 | 2014-01-09 | Thomas F. Gustafson | Informed Traveler Program and Application |
| JP2014010493A (en) * | 2012-06-27 | 2014-01-20 | Koito Electric Industries Ltd | Bus stop information service system |
| CN105096584A (en) * | 2014-05-06 | 2015-11-25 | 中兴通讯股份有限公司 | Traffic decision support method, device, and system |
| CN105702035A (en) * | 2016-03-23 | 2016-06-22 | 武汉大学 | Method for evaluating bus taking difficulty degree by using historical bus data |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782694A (en) * | 2019-10-31 | 2020-02-11 | 同济大学 | Internet bus arrival and stop management optimization method |
| CN111091231A (en) * | 2019-11-25 | 2020-05-01 | 珠海格力电器股份有限公司 | Prediction model training method, time prediction method, training device and terminal |
| CN111091231B (en) * | 2019-11-25 | 2022-04-15 | 珠海格力电器股份有限公司 | Prediction model training method, time prediction method, training device and terminal |
| CN111415679A (en) * | 2020-03-25 | 2020-07-14 | Oppo广东移动通信有限公司 | Site identification method, device, terminal and storage medium |
| CN112256967A (en) * | 2020-10-27 | 2021-01-22 | 北京字跳网络技术有限公司 | Travel recommendation method, device, equipment and storage medium |
| CN114299746A (en) * | 2021-12-30 | 2022-04-08 | 武汉长飞智慧网络技术有限公司 | Image recognition-based bus scheduling method, device, equipment and medium |
| CN115588298A (en) * | 2022-10-28 | 2023-01-10 | 广州地铁集团有限公司 | Urban rail passenger flow broadcasting induction method based on machine vision |
| CN115588298B (en) * | 2022-10-28 | 2023-12-29 | 广州地铁集团有限公司 | Urban rail passenger flow broadcasting induction method based on machine vision |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110245377B (en) | 2023-08-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110245377A (en) | A kind of travel plan recommended method and recommender system | |
| CN109308546B (en) | Method and system for predicting bus trip get-off station of passenger | |
| CN109034566B (en) | A kind of intelligent dispatching method and device based on passenger flow above and below bus station | |
| CN104064028B (en) | Based on public transport arrival time Forecasting Methodology and the system of multiple information data | |
| CN104766473B (en) | Transit trip feature extracting method based on multi-mode public transport Data Matching | |
| CN108629323A (en) | A kind of integrated providing method of scenic spot tourist chain type trip | |
| CN102324128B (en) | Method for predicting OD (Origin-Destination) passenger flow among bus stations on basis of IC (Integrated Circuit)-card record and device | |
| CN108151756A (en) | The travel time computation method and device of a kind of urban public tranlport system | |
| CN110472999B (en) | Passenger flow mode analysis method and device based on subway and shared bicycle data | |
| CN107545320A (en) | A kind of urban track traffic passenger paths planning method and system based on graph theory | |
| CN109102114B (en) | Bus trip getting-off station estimation method based on data fusion | |
| US20230221133A1 (en) | Methods for planning garbage cleaning route in smart cities and internet of things systems thereof | |
| CN105844229A (en) | Method and system for calculating passenger crowdedness degree | |
| CN112347864A (en) | Method, device, equipment and system for sensing and inducing rail transit passenger flow | |
| CN112700473B (en) | Carriage congestion degree judging system based on image recognition | |
| CN106650976A (en) | Travel analysis and forecasting method and system, and travel analysis and forecasting method and system based on IC card | |
| CN108960133A (en) | Method, electronic equipment, system and the storage medium of ridership monitoring | |
| CN109446881A (en) | A kind of express highway section Traffic State Detection Method based on isomeric data | |
| CN110533214A (en) | A kind of subway passenger flow Forecasting Approach for Short-term based on XGBoost algorithm | |
| CN110490103A (en) | Track traffic for passenger flow Density Detection method and device | |
| CN113971893A (en) | Parking space recommendation method and device and storage medium | |
| CN117576943A (en) | Intelligent analysis management system for vehicle information of parking lot | |
| CN110263783A (en) | Multiple features charging addressing analysis of Influential Factors method and system based on deep learning | |
| CN110046535B (en) | Intelligent travel time prediction system, method and storage medium based on machine learning | |
| CN113542364B (en) | Bus and network appointment linkage intelligent traffic system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |