CN108197739A - A kind of urban track traffic ridership Forecasting Methodology - Google Patents
A kind of urban track traffic ridership Forecasting Methodology Download PDFInfo
- Publication number
- CN108197739A CN108197739A CN201711479681.4A CN201711479681A CN108197739A CN 108197739 A CN108197739 A CN 108197739A CN 201711479681 A CN201711479681 A CN 201711479681A CN 108197739 A CN108197739 A CN 108197739A
- Authority
- CN
- China
- Prior art keywords
- ridership
- prediction
- flow
- section
- passengers
- 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 43
- 238000012549 training Methods 0.000 claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000013528 artificial neural network Methods 0.000 claims description 29
- 230000004913 activation Effects 0.000 claims description 21
- 238000013527 convolutional neural network Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 8
- 230000009467 reduction Effects 0.000 claims description 4
- 230000036962 time dependent Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 15
- 238000004364 calculation method Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000011160 research Methods 0.000 description 6
- 230000006399 behavior Effects 0.000 description 5
- 239000000284 extract Substances 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000001131 transforming effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000006854 communication Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (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 present invention provides a kind of urban track traffic ridership Forecasting Methodology, including:Current slot statistics OD distribution matrixs based on target track traffic route, it is established using according to forecast demand, and the ridership prediction model of the history OD distribution matrixs training acquisition using the target track traffic route, the target track traffic route specifies the volume of the flow of passengers in section in prediction specified time section.The present invention can effectively simplify the pre- flow gauge of ridership, improve calculating speed and computational accuracy, and the management and scheduling for rationally progress traffic in time provide strong support.
Description
Technical field
The present invention relates to field of artificial intelligence, are predicted more particularly, to a kind of urban track traffic ridership
Method.
Background technology
The advantages such as rail traffic is saved with its land used, capacity is big, run time is stable, safety and environmental protection are increasingly becoming city friendship
The logical hot spot built.A large amount of research, such as Europe are done in terms of the research and development of Rail Transit System in the countries and regions such as Europe, the United States, day
The DRIVE systems in continent, the TRAVTEK systems in the U.S., VICS systems of Japan etc..These systems are analyzed by vehicle-mounted apparatus for deivation
The road grid traffic information of real-time change, calculates best driving path, and rational management vehicle reaches equilibrium assignment road network wagon flow, carries
The purpose of high conevying efficiency.
In the research of traffic information theory of distribution, development of mathematic programming methods, Graph-theoretical Approach and computer technology etc.,
Development and application for rational Dynamic Traffic Assignment Model provide solid foundation.But due to urban track traffic for passenger flow point
There is apparent difference in research object, Consideration, transfer problem etc. with being distributed with road traffic flow, it is impossible to directly by these
Achievement in research is applied in rail traffic, and needs to be analyzed with reference to rail traffic characteristic.
The country is directed to the research of urban track traffic for passenger flow assignment problem mainly in terms of urban road traffic flow distribution,
The research carried out using traveler housing choice behavior as core.On the basis of the traffic housing choice behavior of analysis passenger, construction track is handed over
The assignment of traffic model and optimization algorithm of logical transit network.Meanwhile based on user equilibrium principle, establish urban mass transit network
Volume of the flow of passengers equilibrium portioning model.These methods may be used in track traffic for passenger flow distribution.
The basic thought of existing Flow Prediction in Urban Mass Transit is:Start with from passenger flow OD distribution matrixs, according to traveler
Travel behaviour always select travel time shortest principle, by Trip Assignment Model simulation traveler in-orbit road traffic programme road
Online distribution.By comparing traveler whether using gap of the Rail Transit System on the travel time, determine to use
The OD amounts of track traffic for passenger flow, and assign it on target track traffic route, obtain target track traffic route website amount
With the line section volume of the flow of passengers.
Traveler often has many influence factors when selecting trip mode and trip route, and wherein principal element has
Row time, trip distance, expense and line arrangement etc..These factors are converted into the broad sense travel time in practical applications.
Traditional passenger flow forecast method needs different target routes to build different models, including right
Trip distance, expense, line arrangement etc. are predicted respectively, and then correct prediction of original model to the volume of the flow of passengers.Therefore,
Model buildings heavy workload, prediction model reusability is low, is unfavorable for the timely management and scheduling of urban track traffic.
Invention content
In order to overcome the above problem or solve the above problems at least partly, the present invention provides a kind of urban track traffic
Ridership Forecasting Methodology effectively to simplify the pre- flow gauge of ridership, improves calculating speed and computational accuracy, to close in time
Reason carries out the management of traffic and scheduling provides strong support.
The present invention provides a kind of urban track traffic ridership Forecasting Methodology, including:Based on target track traffic route
Current slot statistics OD distribution matrixs, using according to OD distribution matrixs architectural characteristic and ridership data characteristic foundation,
And the ridership prediction model of the history OD distribution matrixs training acquisition using the target track traffic route, prediction are specified
The target track traffic route specifies the volume of the flow of passengers in section in period.
Further, the ridership prediction model, the target track traffic in prediction specified time section are being utilized
Before route specifies the step of volume of the flow of passengers in section, the method further includes:S01, the two-dimensional structure based on OD distribution matrixs are special
Property and ridership time-dependent behavior, build CNN-LSTM structures initial ridership prediction model;S02 is obtained
The target track traffic route specifies the history OD distribution matrixs of historical time section, and by the history OD distribution matrixs on time
Between sequentially pre-processed, obtain training sample;S03 predicts mould using the training sample training initial ridership
Type obtains the ridership prediction model.
Wherein, OD distribution matrixs are counted based on the current slot, the target track is handed in prediction specified time section
Access line specifies the step of volume of the flow of passengers in section to further comprise:By current slot statistics OD distribution matrix inputs CNN-
The ridership prediction model of LSTM structures, and it is special to pass through network front end CNN convolutional neural networks extraction current slot passenger flow
Sign amount;Based on the current slot passenger flow characteristic quantity, the target track in LSTM neural network prediction specified times section is utilized
Road traffic route specifies the volume of the flow of passengers in section.
Wherein, the ridership prediction model includes input layer, output layer and hidden layer, and the input layer represents different
Each section volume of the flow of passengers at moment, the output layer represent the volume of the flow of passengers of different moments each section after prediction, and the hidden layer is
LSTM layers, gradually corrected for the error amount according to desired output and reality output so that the reality output follow it is described
Desired output.
Further, OD distribution matrixs are being counted based on the current slot, is utilizing the ridership prediction model
Before carrying out ridership prediction, the method further includes:The data of the section volume of the flow of passengers are pre-processed;Wherein, it presses
The section volume of the flow of passengers is numbered, and be grouped according to predicted time sequence according to time sequencing;Each website is set as
One section, every group of inside is a matrix, and the volume of the flow of passengers is counted every setting time, forms input matrix.
Further, the method further includes:The input matrix through hidden layer LSTM is calculated and is swashed through activation primitive
Processing living;By the input matrix of neural network after invalid data is rejected, input neural network is integrated, obtains the defeated of overall network
Go out.
Wherein, the output of the overall network includes the intermediate output at current time and the prediction of overall network exports;Phase
It answers, the method further includes:According to the passenger flow data of input prediction, prediction output is carried out using subsequent time passenger flow data
Costing bio disturbance, and using backpropagation, to network, each weight carries out gradient calculating, optimizes each weight parameter.
Further, the method further includes:By setting the maximum length for inputting dimension, limiting input matrix, wherein
The use 0 for being unsatisfactory for maximum length is filled, and more than maximum length block.
Further, the method further includes:Pretreated data are filled into the matrix of 299*299, utilize CNN nets
Network carries out feature extraction, and the data flattening after dimensionality reduction is inputted LSTM layers, to accelerate calculating speed.
A kind of urban track traffic ridership Forecasting Methodology provided by the invention, can based on neural network prediction model
It under the support of OD networks, to prejudge Trip distribution, can effectively simplify the pre- flow gauge of ridership, improve calculating speed
And computational accuracy, management and scheduling for rationally progress traffic in time provide strong support.
Description of the drawings
Fig. 1 is a kind of flow chart for establishing rail traffic ridership prediction model of the embodiment of the present invention;
Fig. 2 is a kind of structure diagram of the Passenger flow forecast model based on neural network of the embodiment of the present invention;
Fig. 3 is a kind of structure diagram of the LSTM layers of ridership prediction model of the embodiment of the present invention;
Fig. 4 is a kind of urban track traffic ridership prediction side based on CNN-LSTM neural networks of the embodiment of the present invention
The flow chart of method;
Fig. 5 is a kind of structure diagram of the preposition CNN networks of ridership prediction model of the embodiment of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, is clearly and completely described the technical solution in the present invention, it is clear that described embodiment is one of the present invention
Divide embodiment, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making
All other embodiments obtained under the premise of creative work, shall fall within the protection scope of the present invention.
As one embodiment of the embodiment of the present invention, the present embodiment provides a kind of predictions of urban track traffic ridership
Method, including:Current slot statistics OD distribution matrixs based on target track traffic route, using according to OD distribution matrix knots
Structure characteristic and ridership data characteristic are established, and are obtained using the history OD distribution matrixs training of the target track traffic route
The ridership prediction model taken, the target track traffic route specifies the volume of the flow of passengers in section in prediction specified time section.
It is to be understood that the present embodiment can extract the characteristic of useful feature using neural network from complex data, and
Although each traveler is in Each point in time, different trip distances has different selections, this is discrete, irregular number
According to, but consider using entire trip crowd as an entirety, individual difference can be smoothed, and then a kind of overall rule are presented
Rule.
Simultaneously, it is contemplated that important component of the rail traffic as entire Traffic Systems, according to rail traffic need
Passenger traffic volume is predicted, and then corresponding train table of arranging, the present embodiment are started with from passenger flow OD distribution matrixs, prior basis
Rail line history OD distribution matrixs train neural network prediction model, by learning corrective networks parameter, obtain prediction
Precision meets the ridership prediction model of established standards.
After prediction model foundation, according to data prediction foundation, target track traffic route is predicted using prediction model
In give section given time period the volume of the flow of passengers.Wherein data prediction quantifies really according to for input, and the present embodiment takes target
The current slot statistics OD distribution matrixs of rail traffic route, such as the OD statistical data of circuit to be predicted input current month.
I.e. using passenger flow OD matrixes as the input of whole network, the volume of the flow of passengers of certain a road section in extraction a period of time.
In one embodiment, OD distribution matrixs are being counted based on the current slot, it is pre- using the ridership
It surveys before model progress ridership prediction, the method further includes:The data of the section volume of the flow of passengers are pre-processed;Its
In, the section volume of the flow of passengers is numbered sequentially in time, and is grouped according to predicted time sequence;Each website is set
It is set to a section, every group of inside is a matrix, and the volume of the flow of passengers is counted every setting time, forms input matrix.
It is to be understood that the present embodiment first pre-processes the data of the section volume of the flow of passengers.Sequentially in time to multiplying
Volume of the flow of passengers data are numbered, i.e., according to predicted time sequential packet.Every group possesses unique number, and each group of inside is one
Matrix, it is assumed that a shared l passenger station (as soon as section) at regular intervals counts the volume of the flow of passengers, altogether statistical moment
Battle array has n rows 1 to arrange, and is the volume of the flow of passengers of one node of current time per a line.I.e. network inputs are:
X=(x1,x2,...,xl)T;
In formula, vectorial X represents neural network model input vector, x1,x2,...,xlThe 1st to l statistics section is represented respectively
The volume of the flow of passengers of point.
A kind of urban track traffic ridership Forecasting Methodology provided in an embodiment of the present invention, based on neural network prediction mould
Type can prejudge Trip distribution under the support of OD networks, can effectively simplify the pre- flow gauge of ridership, improve meter
Speed and computational accuracy are calculated, the management and scheduling for rationally progress traffic in time provide strong support.
Further, the ridership prediction model, the target track traffic in prediction specified time section are being utilized
Before route specifies the step of volume of the flow of passengers in section, the method further includes process flow as shown in Figure 1, and Fig. 1 is the present invention
A kind of flow chart for establishing rail traffic ridership prediction model of embodiment, including:
S01 based on the two-dimensional structure characteristic of OD distribution matrixs and the time-dependent behavior of ridership, builds CNN-
The initial ridership prediction model of LSTM structures;
S02 obtains the target track traffic route and specifies the history OD distribution matrixs of historical time section, and gone through described
History OD distribution matrixs are pre-processed in chronological order, obtain training sample;
S03 trains the initial ridership prediction model using the training sample, obtains the ridership prediction
Model.
It is to be understood that according to above-described embodiment, before the prediction of ridership is carried out using prediction model, root is first wanted
The prediction model is established according to existence conditions.Step S01 carries out the initialization structure of target nerve network first, it is contemplated that existing item
Part determines the structure and structural parameters of neural network model.It is specifically contemplated that OD matrixes are inputted for two dimension, chooses two-dimensional process god
It is convolutional neural networks through network.And ridership is then the information with time correlation, therefore is using long memory network in short-term
Best, difficult point is what both networks were single use, and input dimension is different, and setting LSTM neural networks are as pre-
Survey core.
In addition, due to OD distribution matrixs be matrix form, LSTM neural network None- identifieds, it is necessary first to OD moments of distribution
Battle array is converted, and extracts characteristic quantity.The present embodiment before LSTM networks by setting preposition CNN convolutional neural networks, to OD
Distribution matrix carries out process of convolution, then handling result input LSTM neural networks are carried out to the prediction of non-flow of guests.
Step S02 is it is to be understood that the present embodiment specifies the history OD moments of distribution of historical time section with rail traffic route
Based on battle array, the prediction of Future Data is carried out.The OD moments of distribution in given historical time section are obtained by certain approach first
Battle array, such as from fall operator and can obtain city underground and go over OD distribution matrixs in certain time.Then to the data of acquisition
It is pre-processed, forms training sample set.It is divided into time first history OD data and time in history OD distribution matrixs rear
History OD data, the time is formerly and posterior OD data into row label to as one group of training sample, distributing.
Step S03 passes through positive calculate it is to be understood that using time first history OD data as the input of neural network
Prediction result is obtained, and passes through prediction result and is compared with the posterior history OD data of time in training data.According to comparing
Modified result network parameter carries out gradually training for neural network.And final prediction result and training data are met into setting and closed
The neural network of system is as final ridership prediction model.
Wherein optional, the ridership prediction model includes input layer, output layer and hidden layer, the input layer table
Show each section volume of the flow of passengers of different moments, the output layer represents the volume of the flow of passengers of different moments each section after prediction, described implicit
Layer is gradually corrected, so that the reality output follows for the error amount according to desired output and reality output for LSTM layers
The desired output.
It is to be understood that as shown in Fig. 2, for a kind of knot of the Passenger flow forecast model based on neural network of the embodiment of the present invention
Structure schematic diagram.Input layer has t node in neural network in figure, represents the passenger flow of different moments respectively, and XiIt then represents a certain
The volume of the flow of passengers of the cross sections at moment.It exports as t node, represents the volume of the flow of passengers of each period, hiRepresent certain after prediction
The volume of the flow of passengers of one moment cross sections.A blocks are the LSTM layers of neural network core in figure.
LSTM layers of structure are as shown in figure 3, the structure of the LSTM layers for a kind of ridership prediction model of the embodiment of the present invention
Schematic diagram.The learning process of neural network is made of working signal forward-propagating and error signal backpropagation.Just
Into communication process, input signal is transmitted to output layer from input layer through hidden layer, if output layer cannot obtain desired output,
Then be transferred to error signal back-propagation process, by the error between the reality output of network and desired output by output terminal by
Layer judges to correct, and makes the reality output of network closer to desired output.
Usually, it in embodiments herein, is used using different activation primitives according to different network structures
Relu and sigmoid functions are as activation primitive.
It is wherein optional, OD distribution matrixs, the target in prediction specified time section are counted based on the current slot
What rail traffic route specified the volume of the flow of passengers in section is further processed step with reference to figure 4, is based on for one kind of the embodiment of the present invention
The flow chart of the urban track traffic ridership Forecasting Methodology of CNN-LSTM neural networks, including:
Current slot statistics OD distribution matrixs are inputted the ridership prediction model of CNN-LSTM structures by S11,
And pass through network front end CNN convolutional neural networks extraction current slot passenger flow characteristic quantity;
S12, based on the current slot passenger flow characteristic quantity, described in LSTM neural network prediction specified times section
Target track traffic route specifies the volume of the flow of passengers in section.
It is to be understood that before the prediction of Future Data is carried out according to current data using LSTM prediction networks, first lead to
It crosses CNN convolutional neural networks and convolution algorithm is carried out to the OD distribution matrixs of input, extract current slot volume of the flow of passengers characteristic,
Obtain the input X=(x that LSTM networks can identify1,x2,...,xl)T。
Then, in network input layer, activation operation is weighted to the data of input according to the following formula:
In formula,Represent input layer weighting output, ωilRepresent input data weights,During i-th layer of t of representing matrix
The input data at quarter, i-th layer, i=1 of i representing matrixes, 2 ..., I, ωclRepresent last moment status data weights,It represents
Last moment input state value,Represent the input layer output after activation, f () represents input layer activation primitive.
That is, the t of above formula is current time, a is inputted for current time, and b is the input that network is inputted after activation primitive,
Activation primitive is typically chosen in sigmoid, wherein, rear end is the output of a upper moment whole network.
In one embodiment, the method further includes:The input matrix is calculated through hidden layer LSTM and through activation
Function activation is handled;By the input matrix of neural network after invalid data is rejected, input neural network is integrated, obtains integral net
The output of network.
It is to be understood that according to the input at current time, the data that network needs are abandoned are calculated, so that new data progress is defeated
Enter, calculation formula is as follows:
In formula, ωiRepresent input gate i φ matrix weights,The input data of i-th layer of t moment of representing matrix, ωcφIt represents
Last moment c matrix weight, sc t-1Represent state parameter of the last moment after activation primitive.
Wherein, above formula carries out discard processing, unwanted data will abandon before, number of the weights for 0-1,1
To pass through completely, 0 is does not pass through completely.After the calculating of this layer, network is by required data.Wherein, s is defeated for last layer
Enter, x is inputted for this layer, w weights, and b is the input after activation primitive, and rear end is still the defeated of last moment whole network here
Go out.
Later, network middle layer is calculated, by the input of neural network after data are abandoned, it would be desirable to which the data of typing are whole
It closes into network, calculation formula is as follows:
In formula, ωicRepresent the weight of ic matrixes,I-th layer of t moment input of representing matrix, sc t-1Represent last moment warp
State parameter after activation primitive.
Wherein, in formula be another set of weight w, before be to carrying out forgetting processing, forgeing and fall some unwanted data, this
In the data newly inputted are weighted cumulative, this time data is added in network, b is last moment output, acFor
By forgeing door not by the result of g functions activation.It is integrated into b by an activation primitive g, g generally takes sigmod
Function.
Finally, the output of whole network is calculated, network output is two parts, and one exports for the intermediate of current timeAnother is exported for the prediction of whole network.Calculation formula is as follows:
In formula, ωiωIt is i ω matrix weights,I-th layer of t moment input of representing matrix, ωcωRepresent c ω matrix weights,
Represent state parameter of the t moment after activation primitive.
Wherein, it needs to calculate the network backend variable i.e. s at next moment, the specific side for using weighted accumulation
Formula.S inputs updated state parameter for last moment, and x is that the new data of this moment input is by the output of f transforming function transformation functions
B, f transforming function transformation function generally remove tanh functions.
Current time prediction, which exports, is:
In formula, htRepresent neural network forecast output,Represent that network output node layer exports after activation,Represent t moment
State parameter after activation primitive.
Wherein, the output s after calculating and s of the update after f function activates are subjected to matrix multiple.
Wherein optionally, it is defeated to include the intermediate prediction exported with overall network at current time for the output of the overall network
Go out;Correspondingly, the method further includes:It is defeated to predicting using subsequent time passenger flow data according to the passenger flow data of input prediction
Go out and carry out costing bio disturbance, and each weight carries out gradient calculating to network using backpropagation, optimizes each weight parameter.
It is to be understood that by above-mentioned calculating, can obtain according to input X (t)=(x1,x2,...,xl)TBy hiding
Layer calculate and by activation primitive treated output, that is, the passenger flow data H (t) predicted=(h1,h2,...,hm)T.Wherein, to
It measures H and represents hidden layer output vector, h1,h2,...,hmFollowing 1st to the m predicted time section volume of the flow of passengers of prediction is represented respectively.
Then, costing bio disturbance is carried out, and utilize to current predictive output H (t) using subsequent time input data X (t+1)
Backpropagation carries out each weight gradient calculating, and then optimizes each weight parameter.Wherein, label output is Y=(y1,y2,...,
ym)T, the value of expression t moment X (t+1).
According to the value of label and neural network forecast value, loss function can be calculated by cross entropy:
In formula, Γ (x, y) represents loss function, and y represents label value, k representing matrix dimensions, and x represents network output valve.
Above formula is cross entropy calculation formula, and it is prediction number that the deviation y of prediction data and real data, which is real data x,
According to.
It can be reversed using loss function and propagate the weight for correcting each matrix.Above 4 groups calculated after gradient such as
Under:
Above formula is ask partial derivative of the loss function to each weight respectively, and the vector as a result formed is gradient vector, and x represents net
Network exports, and y represents label value.
The basic calculating formula of network is as described above, so as to reach prediction purpose.
Further, the method further includes:By setting the maximum length for inputting dimension, limiting input matrix, wherein
The use 0 for being unsatisfactory for maximum length is filled, and more than maximum length block.
It is to be understood that whole realize needs to handle data, because of the data of input, length differs after treatment,
And it is unified that LSTM networks, which are needed in the input dimension of each dimension, therefore needs exist for the maximum length of setting input
Max_len, the use 0 for being unsatisfactory for maximum length are filled, and more than maximum length block.
Further, the method further includes:Pretreated data are filled into the matrix of 299*299, utilize CNN nets
Network carries out feature extraction, and the data flattening after dimensionality reduction is inputted LSTM layers, to accelerate calculating speed.
It is to be understood that accelerate data in the processing speed of LSTM networks, pretreated data are filled into 299*
299 matrix carries out feature extraction, and the data flattening after dimensionality reduction is inputted into LSTM layers using CNN networks, whole to accelerate
A network calculations speed.
Wherein, the structure of CNN networks is as shown in figure 5, for a kind of preposition CNN of ridership prediction model of the embodiment of the present invention
The structure diagram of network.Input is inputted as the matrix after filling, size 299*299;Later by convolution, most
Great Chiization, convolution and maximum pondization processing, flattening, the input data as LSTM networks are carried out using full articulamentum.
Wherein, in one embodiment, the matrix of convolutional layer selection 3*3, internal weight meet (0,1) normal distribution, fill out
It fills for ' same ';Pond turns to the matrix of 2*2, step-length 4.
In another embodiment, data after treatment are two-dimensional matrix, and each group each for period
The section volume of the flow of passengers, common n groups, i.e., common n period.
Later, two-dimensional matrix is transformed to three-dimensional matrice, form is as follows:[samples, time steps, features].
Wherein, features takes 1, time steps to take the maximum length max_len blocked.One hot subsequently are carried out to label Y
Enconde processing, is input to LSTM networks, and whole network is 32 layers.Loss functions select ' categorical
Crossentropy ', optimum choice ' adam ' select exercise wheel number according to the size of data volume.
Finally, the output by LSTM network operations [batch, 32] thus would know that each node when 32 following
Between in the period volume of the flow of passengers variation predicted value.
According to the predicted value that network calculations obtain, train operation company can more accurately arrange the train of different routes
Cloth, driver, which have holidays by turns etc., reasonably to be arranged.Change the simple distribution mode for increasing train arrangement according to festivals or holidays in the past, play
The efficiently accurately method of salary distribution, carries out daily arrangement for operator and is to provide important decision support.
Finally it should be noted that:The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although
The present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that:It still can be right
Technical solution recorded in foregoing embodiments modifies or carries out equivalent replacement to which part technical characteristic;And this
A little modifications are replaced, the spirit and model of various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (9)
1. a kind of urban track traffic ridership Forecasting Methodology, which is characterized in that including:
Current slot statistics OD distribution matrixs based on target track traffic route, using according to OD distribution matrix architectural characteristics
It is established with ridership data characteristic, and trains what is obtained to multiply using the history OD distribution matrixs of the target track traffic route
Passenger flow forecast model, the target track traffic route specifies the volume of the flow of passengers in section in prediction specified time section.
2. according to the method described in claim 1, it is characterized in that, utilizing the ridership prediction model, prediction is specified
Before the step of target track traffic route specifies the volume of the flow of passengers in section in period, further include:
S01 based on the two-dimensional structure characteristic of OD distribution matrixs and the time-dependent behavior of ridership, builds CNN-LSTM
The initial ridership prediction model of structure;
S02, obtains the history OD distribution matrixs that the target track traffic route specifies historical time section, and by the history OD
Distribution matrix is pre-processed in chronological order, obtains training sample;
S03 trains the initial ridership prediction model using the training sample, obtains the ridership prediction mould
Type.
3. according to the method described in claim 2, it is characterized in that, OD distribution matrixs are counted based on the current slot, in advance
The step of target track traffic route specifies the volume of the flow of passengers in section in specified time section is surveyed to further comprise:
By the ridership prediction model of current slot statistics OD distribution matrix input CNN-LSTM structures, and pass through net
Network front end CNN convolutional neural networks extraction current slot passenger flow characteristic quantity;
Based on the current slot passenger flow characteristic quantity, the target track in LSTM neural network prediction specified times section is utilized
Traffic route specifies the volume of the flow of passengers in section.
4. according to the method described in claim 2, it is characterized in that, the ridership prediction model includes input layer, output
Layer and hidden layer, the input layer represent each section volume of the flow of passengers of different moments, and the output layer represents different moments after prediction
The volume of the flow of passengers of each section, the hidden layer are LSTM layers, are gradually repaiied for error amount according to desired output and reality output
Just, so that the reality output follows the desired output.
5. according to the method described in claim 4, it is characterized in that, based on the current slot count OD distribution matrixs,
Before carrying out ridership prediction using the ridership prediction model, further include:To the data of the section volume of the flow of passengers into
Row pretreatment;
Wherein, the section volume of the flow of passengers is numbered sequentially in time, and is grouped according to predicted time sequence;Each
Website is set as a section, and every group of inside is a matrix, and the volume of the flow of passengers is counted every setting time, forms input square
Battle array.
6. it according to the method described in claim 5, it is characterized in that, further includes:
The input matrix through hidden layer LSTM is calculated and is handled through activation primitive activation;
By the input matrix of neural network after invalid data is rejected, input neural network is integrated, obtains the output of overall network.
7. according to the method described in claim 6, it is characterized in that, the output of the overall network includes the centre at current time
Output and the prediction of overall network export;
Correspondingly, the method further includes:
According to the passenger flow data of input prediction, prediction is exported using subsequent time passenger flow data and carries out costing bio disturbance, and utilize
To network, each weight carries out gradient calculating for backpropagation, optimizes each weight parameter.
8. it according to the method described in claim 4, it is characterized in that, further includes:
By the maximum input dimension of setting, the length of input matrix is limited, wherein the use 0 for being unsatisfactory for maximum length is filled,
More than maximum length block.
9. it according to the method described in claim 5, it is characterized in that, further includes:
Pretreated data are filled into the matrix of 299*299, feature extraction is carried out, and will be after dimensionality reduction using CNN networks
Data flattening inputs LSTM layers, to accelerate calculating speed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711479681.4A CN108197739B (en) | 2017-12-29 | 2017-12-29 | Urban rail transit passenger flow prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711479681.4A CN108197739B (en) | 2017-12-29 | 2017-12-29 | Urban rail transit passenger flow prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108197739A true CN108197739A (en) | 2018-06-22 |
CN108197739B CN108197739B (en) | 2021-03-16 |
Family
ID=62586653
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711479681.4A Active CN108197739B (en) | 2017-12-29 | 2017-12-29 | Urban rail transit passenger flow prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108197739B (en) |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108960532A (en) * | 2018-08-01 | 2018-12-07 | 北京交通大学 | A kind of real-time passenger flow status predication in station and early warning system and method |
CN109120463A (en) * | 2018-10-15 | 2019-01-01 | 新华三大数据技术有限公司 | Method for predicting and device |
CN109166317A (en) * | 2018-10-29 | 2019-01-08 | 东北林业大学 | Method is determined by the time based on the urban transportation path of state feature |
CN109214584A (en) * | 2018-09-21 | 2019-01-15 | 北京百度网讯科技有限公司 | Method and apparatus for passenger flow forecast amount |
CN109583656A (en) * | 2018-12-06 | 2019-04-05 | 重庆邮电大学 | Passenger Flow in Urban Rail Transit prediction technique based on A-LSTM |
CN109886444A (en) * | 2018-12-03 | 2019-06-14 | 深圳市北斗智能科技有限公司 | A kind of traffic passenger flow forecasting, device, equipment and storage medium in short-term |
CN110060471A (en) * | 2019-04-01 | 2019-07-26 | 长安大学 | A kind of vehicle OD stream prediction model construction method and vehicle OD flow prediction technique |
CN110163409A (en) * | 2019-04-08 | 2019-08-23 | 华中科技大学 | A kind of convolutional neural networks dispatching method applied to displacement Flow Shop |
CN110298486A (en) * | 2019-05-29 | 2019-10-01 | 成都理工大学 | A kind of track traffic for passenger flow amount prediction technique based on convolutional neural networks |
CN110443657A (en) * | 2019-08-19 | 2019-11-12 | 泰康保险集团股份有限公司 | Customer traffic data processing method, device, electronic equipment and readable medium |
CN110443422A (en) * | 2019-08-05 | 2019-11-12 | 北京交通大学 | Urban track traffic OD passenger flow forecasting based on OD Attraction Degree |
CN110472800A (en) * | 2019-08-23 | 2019-11-19 | 山东浪潮通软信息科技有限公司 | A kind of machine tool method for predicting residual useful life based on LSTM+CNN |
CN110796301A (en) * | 2019-10-23 | 2020-02-14 | 广东岭南通股份有限公司 | Passenger flow prediction method and device based on IC card data |
CN110852476A (en) * | 2019-09-29 | 2020-02-28 | 华东理工大学 | Passenger flow prediction method and device, computer equipment and storage medium |
CN110969275A (en) * | 2018-09-30 | 2020-04-07 | 杭州海康威视数字技术股份有限公司 | Traffic flow prediction method and device, readable storage medium and electronic device |
CN111027202A (en) * | 2019-12-04 | 2020-04-17 | 北京软通智城科技有限公司 | Method, device and equipment for predicting digital city and storage medium |
CN111582605A (en) * | 2020-05-21 | 2020-08-25 | Oppo广东移动通信有限公司 | Method and device for predicting destination site, electronic equipment and storage medium |
CN111626497A (en) * | 2020-05-25 | 2020-09-04 | 日立楼宇技术(广州)有限公司 | People flow prediction method, device, equipment and storage medium |
CN112001548A (en) * | 2020-08-25 | 2020-11-27 | 北京交通大学 | OD passenger flow prediction method based on deep learning |
CN112215408A (en) * | 2020-09-24 | 2021-01-12 | 交控科技股份有限公司 | Rail transit passenger flow volume prediction method and device |
CN112232607A (en) * | 2020-12-16 | 2021-01-15 | 成都四方伟业软件股份有限公司 | Subway passenger flow volume prediction method and device |
CN112508305A (en) * | 2019-12-29 | 2021-03-16 | 山西大学 | Public place entrance pedestrian flow prediction method based on LSTM |
CN112561128A (en) * | 2020-11-27 | 2021-03-26 | 武汉理工大学 | Method for predicting daily passenger capacity of conventional buses for future urban rail transit transfer |
CN112801377A (en) * | 2021-01-29 | 2021-05-14 | 腾讯大地通途(北京)科技有限公司 | Object estimation method, device, equipment and storage medium |
CN112949931A (en) * | 2021-03-19 | 2021-06-11 | 北京交通大学 | Method and device for predicting charging station data with hybrid data drive and model |
CN113537580A (en) * | 2021-06-28 | 2021-10-22 | 中科领航智能科技(苏州)有限公司 | Public transport passenger flow prediction method and system based on adaptive graph learning |
CN114463978A (en) * | 2022-02-10 | 2022-05-10 | 深圳明弘电子科技有限公司 | Data monitoring method based on rail transit information processing terminal |
CN114841415A (en) * | 2022-04-12 | 2022-08-02 | 西南交通大学 | Urban rail transit passenger flow prediction and multistage transportation organization method during large-scale activities |
CN114842641A (en) * | 2022-03-11 | 2022-08-02 | 华设设计集团股份有限公司 | Provincial-domain-oriented multi-mode chain type traffic distribution method |
CN114912683A (en) * | 2022-05-13 | 2022-08-16 | 中铁第六勘察设计院集团有限公司 | Intelligent urban rail transit abnormal large passenger flow prediction system and method |
CN116050673A (en) * | 2023-03-31 | 2023-05-02 | 深圳市城市交通规划设计研究中心股份有限公司 | Urban public transport passenger flow short-time prediction method based on CNN-BiLSTM |
-
2017
- 2017-12-29 CN CN201711479681.4A patent/CN108197739B/en active Active
Non-Patent Citations (1)
Title |
---|
王科等: "基于神经网络的高速公路出入口OD矩阵估计方法研究", 《交通与计算机》 * |
Cited By (51)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108960532A (en) * | 2018-08-01 | 2018-12-07 | 北京交通大学 | A kind of real-time passenger flow status predication in station and early warning system and method |
CN109214584A (en) * | 2018-09-21 | 2019-01-15 | 北京百度网讯科技有限公司 | Method and apparatus for passenger flow forecast amount |
US11526748B2 (en) | 2018-09-21 | 2022-12-13 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method and apparatus for predicting passenger flow |
CN109214584B (en) * | 2018-09-21 | 2022-02-08 | 北京百度网讯科技有限公司 | Method and device for predicting passenger flow |
CN110969275A (en) * | 2018-09-30 | 2020-04-07 | 杭州海康威视数字技术股份有限公司 | Traffic flow prediction method and device, readable storage medium and electronic device |
CN110969275B (en) * | 2018-09-30 | 2024-01-23 | 杭州海康威视数字技术股份有限公司 | Traffic flow prediction method and device, readable storage medium and electronic equipment |
CN109120463A (en) * | 2018-10-15 | 2019-01-01 | 新华三大数据技术有限公司 | Method for predicting and device |
CN109166317A (en) * | 2018-10-29 | 2019-01-08 | 东北林业大学 | Method is determined by the time based on the urban transportation path of state feature |
CN109886444A (en) * | 2018-12-03 | 2019-06-14 | 深圳市北斗智能科技有限公司 | A kind of traffic passenger flow forecasting, device, equipment and storage medium in short-term |
CN109583656A (en) * | 2018-12-06 | 2019-04-05 | 重庆邮电大学 | Passenger Flow in Urban Rail Transit prediction technique based on A-LSTM |
CN109583656B (en) * | 2018-12-06 | 2022-05-10 | 重庆邮电大学 | Urban rail transit passenger flow prediction method based on A-LSTM |
CN110060471A (en) * | 2019-04-01 | 2019-07-26 | 长安大学 | A kind of vehicle OD stream prediction model construction method and vehicle OD flow prediction technique |
CN110163409B (en) * | 2019-04-08 | 2021-05-18 | 华中科技大学 | Convolutional neural network scheduling method applied to replacement flow shop |
CN110163409A (en) * | 2019-04-08 | 2019-08-23 | 华中科技大学 | A kind of convolutional neural networks dispatching method applied to displacement Flow Shop |
CN110298486A (en) * | 2019-05-29 | 2019-10-01 | 成都理工大学 | A kind of track traffic for passenger flow amount prediction technique based on convolutional neural networks |
CN110443422A (en) * | 2019-08-05 | 2019-11-12 | 北京交通大学 | Urban track traffic OD passenger flow forecasting based on OD Attraction Degree |
CN110443422B (en) * | 2019-08-05 | 2021-11-19 | 北京交通大学 | OD attraction degree-based urban rail transit OD passenger flow prediction method |
CN110443657B (en) * | 2019-08-19 | 2022-03-18 | 泰康保险集团股份有限公司 | Client flow data processing method and device, electronic equipment and readable medium |
CN110443657A (en) * | 2019-08-19 | 2019-11-12 | 泰康保险集团股份有限公司 | Customer traffic data processing method, device, electronic equipment and readable medium |
CN110472800A (en) * | 2019-08-23 | 2019-11-19 | 山东浪潮通软信息科技有限公司 | A kind of machine tool method for predicting residual useful life based on LSTM+CNN |
CN110852476A (en) * | 2019-09-29 | 2020-02-28 | 华东理工大学 | Passenger flow prediction method and device, computer equipment and storage medium |
CN110796301B (en) * | 2019-10-23 | 2022-11-11 | 广东岭南通股份有限公司 | Passenger flow prediction method and device based on IC card data |
CN110796301A (en) * | 2019-10-23 | 2020-02-14 | 广东岭南通股份有限公司 | Passenger flow prediction method and device based on IC card data |
CN111027202A (en) * | 2019-12-04 | 2020-04-17 | 北京软通智城科技有限公司 | Method, device and equipment for predicting digital city and storage medium |
CN111027202B (en) * | 2019-12-04 | 2023-12-15 | 北京软通绿城科技有限公司 | Digital city prediction method, device, equipment and storage medium |
CN112508305A (en) * | 2019-12-29 | 2021-03-16 | 山西大学 | Public place entrance pedestrian flow prediction method based on LSTM |
CN111582605A (en) * | 2020-05-21 | 2020-08-25 | Oppo广东移动通信有限公司 | Method and device for predicting destination site, electronic equipment and storage medium |
CN111582605B (en) * | 2020-05-21 | 2023-09-12 | Oppo广东移动通信有限公司 | Method and device for predicting destination site, electronic equipment and storage medium |
CN111626497A (en) * | 2020-05-25 | 2020-09-04 | 日立楼宇技术(广州)有限公司 | People flow prediction method, device, equipment and storage medium |
CN112001548B (en) * | 2020-08-25 | 2023-10-20 | 北京交通大学 | OD passenger flow prediction method based on deep learning |
CN112001548A (en) * | 2020-08-25 | 2020-11-27 | 北京交通大学 | OD passenger flow prediction method based on deep learning |
CN112215408A (en) * | 2020-09-24 | 2021-01-12 | 交控科技股份有限公司 | Rail transit passenger flow volume prediction method and device |
CN112561128A (en) * | 2020-11-27 | 2021-03-26 | 武汉理工大学 | Method for predicting daily passenger capacity of conventional buses for future urban rail transit transfer |
CN112561128B (en) * | 2020-11-27 | 2022-06-10 | 武汉理工大学 | Method for predicting daily passenger capacity of conventional buses for future urban rail transit transfer |
CN112232607A (en) * | 2020-12-16 | 2021-01-15 | 成都四方伟业软件股份有限公司 | Subway passenger flow volume prediction method and device |
CN112232607B (en) * | 2020-12-16 | 2021-03-09 | 成都四方伟业软件股份有限公司 | Subway passenger flow volume prediction method and device |
CN112801377B (en) * | 2021-01-29 | 2023-08-22 | 腾讯大地通途(北京)科技有限公司 | Object estimation method, device, equipment and storage medium |
CN112801377A (en) * | 2021-01-29 | 2021-05-14 | 腾讯大地通途(北京)科技有限公司 | Object estimation method, device, equipment and storage medium |
CN112949931B (en) * | 2021-03-19 | 2024-03-08 | 北京交通大学 | Method and device for predicting charging station data by mixing data driving and models |
CN112949931A (en) * | 2021-03-19 | 2021-06-11 | 北京交通大学 | Method and device for predicting charging station data with hybrid data drive and model |
CN113537580A (en) * | 2021-06-28 | 2021-10-22 | 中科领航智能科技(苏州)有限公司 | Public transport passenger flow prediction method and system based on adaptive graph learning |
CN113537580B (en) * | 2021-06-28 | 2024-04-09 | 中科领航智能科技(苏州)有限公司 | Public transportation passenger flow prediction method and system based on self-adaptive graph learning |
CN114463978A (en) * | 2022-02-10 | 2022-05-10 | 深圳明弘电子科技有限公司 | Data monitoring method based on rail transit information processing terminal |
CN114463978B (en) * | 2022-02-10 | 2024-03-29 | 深圳明弘电子科技有限公司 | Data monitoring method based on track traffic information processing terminal |
CN114842641B (en) * | 2022-03-11 | 2024-02-09 | 华设设计集团股份有限公司 | Multi-mode chain traffic distribution method for province domain |
CN114842641A (en) * | 2022-03-11 | 2022-08-02 | 华设设计集团股份有限公司 | Provincial-domain-oriented multi-mode chain type traffic distribution method |
CN114841415A (en) * | 2022-04-12 | 2022-08-02 | 西南交通大学 | Urban rail transit passenger flow prediction and multistage transportation organization method during large-scale activities |
CN114912683A (en) * | 2022-05-13 | 2022-08-16 | 中铁第六勘察设计院集团有限公司 | Intelligent urban rail transit abnormal large passenger flow prediction system and method |
CN114912683B (en) * | 2022-05-13 | 2024-05-10 | 中铁第六勘察设计院集团有限公司 | System and method for predicting abnormal large passenger flow of smart city rail transit |
CN116050673B (en) * | 2023-03-31 | 2023-08-01 | 深圳市城市交通规划设计研究中心股份有限公司 | Urban public transport passenger flow short-time prediction method based on CNN-BiLSTM |
CN116050673A (en) * | 2023-03-31 | 2023-05-02 | 深圳市城市交通规划设计研究中心股份有限公司 | Urban public transport passenger flow short-time prediction method based on CNN-BiLSTM |
Also Published As
Publication number | Publication date |
---|---|
CN108197739B (en) | 2021-03-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108197739A (en) | A kind of urban track traffic ridership Forecasting Methodology | |
CN109887282B (en) | Road network traffic flow prediction method based on hierarchical timing diagram convolutional network | |
CN111376954B (en) | Train autonomous scheduling method and system | |
CN111369181B (en) | Train autonomous scheduling deep reinforcement learning method and device | |
CN109285346A (en) | A kind of city road net traffic state prediction technique based on key road segment | |
CN109886444A (en) | A kind of traffic passenger flow forecasting, device, equipment and storage medium in short-term | |
CN109272157A (en) | A kind of freeway traffic flow parameter prediction method and system based on gate neural network | |
CN110223517A (en) | Short-term traffic flow forecast method based on temporal correlation | |
CN110517482B (en) | Short-term traffic flow prediction method based on 3D convolutional neural network | |
CN109840628A (en) | A kind of multizone speed prediction method and system in short-term | |
CN106127329A (en) | Order forecast method and device | |
CN106447119A (en) | Short-term traffic flow prediction method and system based on convolutional neural network | |
CN107909206A (en) | A kind of PM2.5 Forecasting Methodologies based on deep structure Recognition with Recurrent Neural Network | |
CN106529820A (en) | Operation index prediction method and system | |
CN106781489A (en) | A kind of road network trend prediction method based on recurrent neural network | |
CN108009632A (en) | Confrontation type space-time big data Forecasting Methodology | |
CN107832913A (en) | The Forecasting Methodology and system to monitoring data trend based on deep learning | |
CN108122048A (en) | A kind of transportation route dispatching method and its system | |
CN108346293A (en) | A kind of arithmetic for real-time traffic flow Forecasting Approach for Short-term | |
CN110599236A (en) | Short-time parking demand prediction method based on GRU model | |
CN103942461A (en) | Water quality parameter prediction method based on online sequential extreme learning machine | |
CN109840639A (en) | A kind of late time forecasting methods of high speed rail train operation | |
CN107392389A (en) | Taxi dispatching processing method based on ARIMA models | |
CN105204472A (en) | Single-piece discrete type production operation scheduling optimization method | |
CN107067076A (en) | A kind of passenger flow forecasting based on time lag NARX neutral nets |
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 |