CN112488422A - Multi-mode travel demand prediction method based on multi-task learning - Google Patents
Multi-mode travel demand prediction method based on multi-task learning Download PDFInfo
- Publication number
- CN112488422A CN112488422A CN202011485155.0A CN202011485155A CN112488422A CN 112488422 A CN112488422 A CN 112488422A CN 202011485155 A CN202011485155 A CN 202011485155A CN 112488422 A CN112488422 A CN 112488422A
- Authority
- CN
- China
- Prior art keywords
- travel
- travel demand
- grid
- time
- mode
- 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 32
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000010586 diagram Methods 0.000 claims description 18
- 238000013527 convolutional neural network Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 7
- 239000000126 substance Substances 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 230000029305 taxis Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- G06Q50/40—
Abstract
The invention discloses a multi-mode travel demand prediction method based on multitask learning, which relates to the technical field of traffic demand prediction and specifically comprises the following steps: collecting non-aggregate travel data of various traffic modes, and preprocessing the non-aggregate travel data; constructing a travel demand grid graph; constructing a time sequence of a local trip demand grid graph, constructing a multi-task learning sample set, and constructing the time sequence of the local trip demand grid graph into the multi-task learning sample set; constructing a travel demand prediction model based on multi-task learning; training the travel demand prediction model based on multi-task learning by using the multi-task learning sample set to obtain a trained travel demand prediction model; and the travel demand prediction of multiple modes can be carried out by utilizing the trained travel demand prediction model. The invention integrates various travel demand prediction tasks and improves the prediction precision of travel demands.
Description
Technical Field
The invention relates to the technical field of traffic demand prediction, in particular to a multi-mode travel demand prediction method based on multi-task learning.
Background
In recent years, novel travel modes such as net appointment cars and shared bicycles are gradually popular with people, and the demand of people for novel traffic resources is increased day by day. Therefore, traffic resources are better scheduled, the operation cost of enterprises is reduced, and the quality of travel service is improved by accurately predicting travel demands, which is more and more important.
In the existing travel demand prediction methods, a part adopts a traditional linear model and a machine learning method, and the other part adopts a deep learning method. Most of the existing methods belong to single-task learning, and a travel demand prediction task integrating multiple traffic modes is not considered.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a multi-mode travel demand prediction method based on multi-task learning.
The invention adopts the following technical scheme for solving the technical problems:
the invention provides a multi-mode travel demand prediction method based on multi-task learning, which comprises the following steps:
dividing the research area by adopting grids to obtain a spatial grid map GI×J，GI×JIn (1) spaceThe row coordinates of the grids are represented by i, the column coordinates are represented by j, the time granularity for counting the number of travel records is set as T, one day is divided into time slices according to the time granularity, each travel record in the standardized travel record data is matched to the corresponding time slice and the corresponding space grid according to the departure time and the field of the latitude and longitude of the departure point, the number of the travel records of each space grid in each time slice is counted and is used as a travel demand, and therefore the travel demand is obtained
And travel demand grid diagram Dt,I×J(ii) a Wherein G isI×JRepresenting a spatial grid comprising I rows and J columns,
representing the travel demand of the spatial grid of the ith row and the jth column in the t time slice, Dt,I×JThe method comprises the steps of representing a travel demand grid diagram of a spatial grid diagram containing I lines and J lines at a t-th time slice, wherein I represents the line number of the spatial grid diagram, and J represents the column number of the spatial grid diagram;
step 3, constructing a time sequence of the local travel demand grid map, which is specifically as follows:
sliding the sliding window of S multiplied by S on the travel demand grid map to enable the central grid of the sliding window to traverse all the spatial grids, and accordingly obtaining a local travel demand grid map LtArranging the local trip demand grid graphs according to a time sequence to obtain a time sequence of the local trip demand grid graphs; wherein L istThe grid diagram of the local travel demand at the t-th time slice is shown, and S is the side length of the sliding window;
step 4, constructing a multi-task learning sample set, and constructing a time sequence of the local trip demand grid graph into a multi-task learning sample set D;
step 5, constructing a travel demand prediction model based on multitask learning, wherein the travel demand prediction model based on the multitask learning comprises N front modules, a sharing module and N rear modules, the front modules comprise M Convolutional Neural Network (CNN) layers, M full-connection layers and long and short time memory neural network (LSTM) layers, the M Convolutional Neural Network (CNN) layers are connected with the M full-connection layers in a one-to-one correspondence mode, the M full-connection layers are respectively connected with the long and short time memory neural network (LSTM) layers, and the sharing module and the rear modules respectively comprise a first full-connection layer, a discarding layer and a second full-connection layer which are sequentially connected in sequence; n is the number of types of transportation modes, and M represents the length of a time sequence of a local travel demand grid graph input into the constructed travel demand prediction model based on the multi-task learning in D;
and 7, utilizing the travel demand prediction model trained in the step 6 to predict the travel demand in multiple modes.
As a further optimization scheme of the multi-mode travel demand prediction method based on multi-task learning, the preprocessing in the step 1 refers to: removing data which are absent from field information in non-ensemble calculated row data and are not in a research area; the non-aggregate trip data in step 1 comprises the following fields: departure time, departure point longitude and latitude; data lacking any one of the fields is regarded as data with incomplete information, and data not in the research area is regarded as redundant data and is removed.
As a further optimization scheme of the multi-mode travel demand prediction method based on multitask learning, in step 2, each travel record in the standardized travel record data is matched to a corresponding time slice, which means that: matching the travel record of the travel time in the tth time slice with the tth time slice, wherein the step of matching the travel record with the time slice is to add a time slice field for the standardized travel record data; in step 2, matching each trip record in the standardized trip record data to a corresponding spatial grid means that: calculating the longitude and latitude of the boundary of the spatial grid, matching the travel records of which the travel longitudes and latitudes are located in the boundary of the spatial grid of the ith row and the jth column to the spatial grid of the ith row and the jth column, and matching the travel records to the spatial grid.
As a further optimization scheme of the multi-mode travel demand prediction method based on multi-task learning, the size I, J of the spatial grid map is 10, 15 or 20, and the time granularity T is 30 min.
As a further optimization scheme of the multi-mode travel demand prediction method based on multitask learning, the side length S of the sliding window in the step 3 is set to be 5, and a local travel demand grid map is selected on the travel demand grid map by using a sliding window of 5 multiplied by 5; and arranging according to the time sequence to obtain a time sequence of the local travel demand grid map.
As a further optimization scheme of the multi-mode travel demand prediction method based on multi-task learning, in step 4, a multi-task learning sample set D ═ D1,D2,…,Dn,…,DNIn which D isnA sample set representing the nth traffic mode, N is an integer greater than or equal to 1 and less than or equal to N, DnIs represented as
Wherein the content of the first and second substances,
the characteristics of the kth sample of the sample set representing the nth mode of transportation,
a grid map showing local travel demand for the nth mode of transportation at time slice (t-M), where M is 1,2,3, …, M,
a label representing the kth sample of the nth mode of transportation,
and the travel demand of the nth transportation mode in the t time slice is shown.
The multi-mode travel demand prediction method based on multi-task learning further optimizes the scheme that N is 2 or 3, and M takes 8.
As a further optimization scheme of the multi-mode travel demand prediction method based on multitask learning, in step 5, the layer-to-layer specific connection relationship of the travel demand prediction model based on multitask learning includes:
step 51, the characteristics of the kth sample of the sample set of the nth transportation mode
Input to the nth front module, the front modules are firstly coupled
Performing convolution and full-connection processing, then performing long-time and short-time memory processing, and taking the output of the nth front module as the travel demand space-time characteristic
Wherein the content of the first and second substances,
representing the travel demand space-time characteristics of the nth traffic mode of the t time slice;
step 52, obtaining the travel demand space-time characteristics
Input to the sharing module, sharing module pair
Carrying out full connection, abandonment and full connection processing to obtain the fusion travel demand characteristic HtWherein H istRepresenting the fusion travel demand characteristics of the t time slice;
step 53, integrating the travel demand characteristics HtInputting all the post-modules, and respectively aligning the post-modules with the HtPerforming full connection, discarding and full connection processing, the output of the nth post-module is
Wherein the content of the first and second substances,
and the travel demand of the nth transportation mode of the t time slice is shown.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
the multi-mode travel demand prediction method based on multi-task learning disclosed by the invention is integrated with multi-mode travel demand prediction tasks, and a travel demand prediction model based on multi-task learning is designed and is divided into three modules: the system comprises a front module, a sharing module and a rear module, wherein the front module is responsible for receiving input characteristics, the sharing module splices the intermediate characteristics of different transportation modes and outputs fusion characteristics, and the rear module receives the fusion characteristics of the front module and outputs travel demands of different modes. The method can predict the travel demands of various traffic modes at the same time, and the prediction effect of the travel demands of various traffic modes is better than that of the travel demand of a single traffic mode by fusing the intermediate characteristics of the travel demands of different traffic modes.
Drawings
FIG. 1 is a schematic diagram of the technical scheme of the invention.
Fig. 2 is a schematic diagram of a front end module of the present invention.
FIG. 3 is a schematic diagram of a shared module of the present invention.
Fig. 4 is a schematic diagram of the back-end module of the present invention.
Fig. 5 is a learning curve presentation diagram of an embodiment of the present invention.
FIG. 6 is a diagram showing the predicted effect of the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
The specific implementation mode of the invention comprises the following steps and contents:
the technical route of the invention is shown in fig. 1, and mainly comprises seven steps of calculating travel data for a non-collection pretreatment, dividing space grids and time slices, calculating the number of travel records, constructing a travel demand grid graph, constructing a multi-task learning sample set, training a multi-task learning model and predicting multi-mode travel demands.
The transportation modes of urban resident traveling mainly comprise buses, subways, network appointment cars, taxis, public bicycles, shared bicycles, private cars and the like. When the number N of the transportation modes is 2, the transportation modes to be studied may be any two of the aforementioned transportation modes; when the number N of transportation modes is 3, the transportation mode to be studied may be any three of the aforementioned transportation modes. In the present embodiment, N is 2, and the transportation modes to be studied are taxis and public bikes.
1) Preprocessing non-aggregate travel data: in the embodiment, taxi order data and public bicycle data in suzhou city are selected as research objects, and basic fields of the original data are shown in table 1:
TABLE 1 non-aggregated travel data field
The fStartLON, fStartLAT, fStartTime fields in the taxi data and the LEASETIME, lending longitude, lending latitude fields in the bicycle data are fields that need to be used in this embodiment. In the process of preprocessing the original data, the following operations are mainly performed: and deleting the record containing the missing value, converting the longitude and latitude of the geographic coordinate system into coordinates under a UTM projection coordinate system, and splitting the time field into a month field, a day field, an hour field and a minute field. The preprocessed data fields are shown in table 2:
TABLE 2 processed data fields
| Traffic mode | Post-processing data field |
| Taxi | Mouth，Day，hour，min，x，y |
| Public bicycle | Mouth，Day，hour，min，x，y |
2) Dividing a space grid: the research area of this embodiment is a part of the city area of suzhou ancient city, the position of the suzhou ancient city (x is 271551.6m, y is 3466875.9m) is used as the center, the size of the spatial grid is set to 400m × 400m, and a 10 × 10 spatial grid area is selected as the research area;
3) constructing a multi-task learning sample set;
4) the method comprises the following steps of (1) realizing a multi-task learning prediction model: the example uses a Python language-based pytorech package to write the code of the multi-task learning prediction model, which comprises three modules: the device comprises a front module, a sharing module and a rear module;
the connection relationship between the layers of the front module is shown in fig. 2, the connection relationship between the layers of the shared module is shown in fig. 3, and the connection relationship between the layers of the rear module is shown in fig. 4;
5) training a model: selecting the root mean square error as a loss function of the model, setting the number of training rounds to be 60, wherein a variation curve of the root mean square error of the model on the data set of the embodiment along with the number of the training rounds is shown in fig. 5;
according to the training curve of the model, when the number of training rounds reaches about 50 times, the loss function tends to be unchanged, so that the model parameters obtained by the 50 th training are used as the optimal parameters, namely the parameters of the optimal model;
6) and (3) realizing prediction: inputting the new data into the demand forecasting model to obtain a forecasting value of the trip demand, and a comparison graph of the forecasting value and the actual value is shown in fig. 6, wherein the abscissa is a label of a time slice, and the ordinate is the order quantity.
The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (8)
1. A multi-mode travel demand prediction method based on multitask learning is characterized by comprising the following steps:
step 1, collecting non-collective travel data of multiple transportation modes, preprocessing the non-collective travel data, and taking the preprocessed non-collective travel data as standardized travel record data; the non-collected travel data comprises departure time and departure point latitude and longitude fields;
step 2, constructing a travel demand grid map; the method comprises the following specific steps:
dividing the research area by adopting grids to obtain a spatial grid map GI×J，GI×JThe row coordinates of the space grid are represented by i, the column coordinates are represented by j, the time granularity for counting the quantity of travel records is set as T, one day is divided into time slices according to the time granularity, each travel record in the standardized travel record data is matched to the corresponding time slice and space grid according to the departure time and the field of the latitude and longitude of the departure point, the quantity of the travel records of each space grid in each time slice is counted and is used as a travel demand, and therefore the travel demand is obtained
And travel demand grid diagram Dt，I×J(ii) a Wherein G isI×JRepresenting a spatial grid comprising I rows and J columns,
representing the travel demand of the spatial grid of the ith row and the jth column in the t time slice, Dt，I×JThe method comprises the steps of representing a travel demand grid diagram of a spatial grid diagram containing I lines and J lines at a t-th time slice, wherein I represents the line number of the spatial grid diagram, and J represents the column number of the spatial grid diagram;
step 3, constructing a time sequence of the local travel demand grid map, which is specifically as follows:
sliding the sliding window of S multiplied by S on the travel demand grid map to enable the central grid of the sliding window to traverse all the spatial grids, and accordingly obtaining a local travel demand grid map LtArranging the local trip demand grid graphs according to a time sequence to obtain a time sequence of the local trip demand grid graphs; wherein L istThe grid diagram of the local travel demand at the t-th time slice is shown, and S is the side length of the sliding window;
step 4, constructing a multi-task learning sample set, and constructing a time sequence of the local trip demand grid graph into a multi-task learning sample set D;
step 5, constructing a travel demand prediction model based on multitask learning, wherein the travel demand prediction model based on the multitask learning comprises N front modules, a sharing module and N rear modules, the front modules comprise M Convolutional Neural Network (CNN) layers, M full-connection layers and long and short time memory neural network (LSTM) layers, the M Convolutional Neural Network (CNN) layers are connected with the M full-connection layers in a one-to-one correspondence mode, the M full-connection layers are respectively connected with the long and short time memory neural network (LSTM) layers, and the sharing module and the rear modules respectively comprise a first full-connection layer, a discarding layer and a second full-connection layer which are sequentially connected in sequence; n is the number of types of transportation modes, and M represents the length of a time sequence of a local travel demand grid graph input into the constructed travel demand prediction model based on the multi-task learning in D;
step 6, training the travel demand prediction model based on the multi-task learning in the step 5 by using the multi-task learning sample set in the step 4 to obtain a trained travel demand prediction model;
and 7, utilizing the travel demand prediction model trained in the step 6 to predict the travel demand in multiple modes.
2. The multi-mode travel demand prediction method based on multitask learning according to claim 1, characterized in that the preprocessing in the step 1 is: removing data which are absent from field information in non-ensemble calculated row data and are not in a research area; the non-aggregate trip data in step 1 comprises the following fields: departure time, departure point longitude and latitude; data lacking any one of the fields is regarded as data with incomplete information, and data not in the research area is regarded as redundant data and is removed.
3. The multi-mode travel demand prediction method based on multitask learning according to claim 1, wherein the step 2 of matching each travel record in the standardized travel record data to a corresponding time slice refers to: matching the travel record of the travel time in the tth time slice with the tth time slice, wherein the step of matching the travel record with the time slice is to add a time slice field for the standardized travel record data; in step 2, matching each trip record in the standardized trip record data to a corresponding spatial grid means that: calculating the longitude and latitude of the boundary of the spatial grid, matching the travel records of which the travel longitudes and latitudes are located in the boundary of the spatial grid of the ith row and the jth column to the spatial grid of the ith row and the jth column, and matching the travel records to the spatial grid.
4. The multi-mode travel demand prediction method based on multitask learning according to claim 1, characterized in that the size I, J of the spatial grid graph is 10 or 15 or 20, and the time granularity T is 30 mmin.
5. The multi-mode travel demand prediction method based on multitask learning according to claim 1, characterized in that the side length S of the sliding window in step 3 is set to 5, and a local travel demand grid map is selected on the travel demand grid map by using a sliding window of 5 × 5; and arranging according to the time sequence to obtain a time sequence of the local travel demand grid map.
6. The multi-mode travel demand prediction method based on multitask learning according to claim 1, characterized in that in step 4, a multitask learning sample set D ═ { D ═ is used1，D2，...，Dn，...，DNIn which D isnA sample set representing the nth traffic mode, N is an integer greater than or equal to 1 and less than or equal to N, DnIs represented as
Wherein the content of the first and second substances,
the characteristics of the kth sample of the sample set representing the nth mode of transportation,
a grid map representing local travel demand for the nth mode of transportation at the (t-M) th time slice, M being 1,2, 3.
A label representing the kth sample of the nth mode of transportation,
and the travel demand of the nth transportation mode in the t time slice is shown.
7. The multi-mode travel demand prediction method based on multitask learning according to claim 1, wherein N is 2 or 3, and M takes a value of 8.
8. The multi-mode travel demand prediction method based on multitask learning according to claim 1, wherein the specific connection relationship between layers of the travel demand prediction model based on multitask learning in the step 5 comprises:
step 51, the characteristics of the kth sample of the sample set of the nth transportation mode
Input to the nth front module, the front modules are firstly coupled
Performing convolution and full-connection processing, then performing long-time and short-time memory processing, and taking the output of the nth front module as the travel demand space-time characteristic
Wherein the content of the first and second substances,
representing the travel demand space-time characteristics of the nth traffic mode of the t time slice;
step 52, obtaining the travel demand space-time characteristics
Input to the sharing module, sharing module pair
Performing full connection, discarding and full connection treatment to obtain a fused productLine demand characteristic HtWherein H istRepresenting the fusion travel demand characteristics of the t time slice;
step 53, integrating the travel demand characteristics HtInputting all the post-modules, and respectively aligning the post-modules with the HtPerforming full connection, discarding and full connection processing, the output of the nth post-module is
Wherein the content of the first and second substances,
and the travel demand of the nth transportation mode of the t time slice is shown.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011485155.0A CN112488422B (en) | 2020-12-16 | 2020-12-16 | Multi-mode travel demand prediction method based on multi-task learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011485155.0A CN112488422B (en) | 2020-12-16 | 2020-12-16 | Multi-mode travel demand prediction method based on multi-task learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112488422A true CN112488422A (en) | 2021-03-12 |
| CN112488422B CN112488422B (en) | 2022-11-18 |
Family
ID=74917992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011485155.0A Active CN112488422B (en) | 2020-12-16 | 2020-12-16 | Multi-mode travel demand prediction method based on multi-task learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112488422B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113837383A (en) * | 2021-10-18 | 2021-12-24 | 中国联合网络通信集团有限公司 | Model training method and device, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633871A (en) * | 2019-09-25 | 2019-12-31 | 大连理工大学 | Regional traffic demand prediction method based on convolution long-term and short-term memory network |
| CN111507762A (en) * | 2020-04-15 | 2020-08-07 | 中国科学院自动化研究所 | Urban taxi demand prediction method based on multi-task co-prediction neural network |
| CN111915081A (en) * | 2020-08-03 | 2020-11-10 | 东北大学秦皇岛分校 | Peak-sensitive travel demand prediction method based on deep learning |
-
2020
- 2020-12-16 CN CN202011485155.0A patent/CN112488422B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633871A (en) * | 2019-09-25 | 2019-12-31 | 大连理工大学 | Regional traffic demand prediction method based on convolution long-term and short-term memory network |
| CN111507762A (en) * | 2020-04-15 | 2020-08-07 | 中国科学院自动化研究所 | Urban taxi demand prediction method based on multi-task co-prediction neural network |
| CN111915081A (en) * | 2020-08-03 | 2020-11-10 | 东北大学秦皇岛分校 | Peak-sensitive travel demand prediction method based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113837383A (en) * | 2021-10-18 | 2021-12-24 | 中国联合网络通信集团有限公司 | Model training method and device, electronic equipment and storage medium |
| CN113837383B (en) * | 2021-10-18 | 2023-06-23 | 中国联合网络通信集团有限公司 | Model training method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112488422B (en) | 2022-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11449908B2 (en) | Method and system for determining transport service fee | |
| AU2019101806A4 (en) | Methods and systems for providing information for an on-demand service | |
| CN111653088B (en) | Vehicle driving quantity prediction model construction method, prediction method and system | |
| CN106448233B (en) | Public bus network timetable cooperative optimization method based on big data | |
| CN105809292B (en) | Bus IC card passenger getting off car website projectional technique | |
| CN113077090B (en) | Passenger flow prediction method, system and computer readable storage medium | |
| Li et al. | Deep learning based parking prediction on cloud platform | |
| CN107563540A (en) | A kind of public transport in short-term based on random forest is got on the bus the Forecasting Methodology of the volume of the flow of passengers | |
| CN112017436B (en) | Method and system for predicting urban traffic travel time | |
| CN111461426A (en) | High-precision travel time length prediction method based on deep learning | |
| CN112488422B (en) | Multi-mode travel demand prediction method based on multi-task learning | |
| CN107657006B (en) | Public bicycle IC card and subway IC card matching method based on time-space characteristics | |
| Laifa et al. | Train delay prediction in Tunisian railway through LightGBM model | |
| CN116153125A (en) | Space dimension bus arrival time prediction method based on IDA-RNN model | |
| CN114202120A (en) | Urban traffic travel time prediction method aiming at multi-source heterogeneous data | |
| CN111199247A (en) | Bus operation simulation method | |
| CN111985731B (en) | Method and system for predicting number of people at urban public transport station | |
| Lin et al. | Insights into Travel Pattern Analysis and Demand Prediction: A Data-Driven Approach in Bike-Sharing Systems | |
| CN115965163A (en) | Rail transit short-time passenger flow prediction method for generating countermeasures to loss based on space-time diagram | |
| CN113222299B (en) | Intelligent vehicle scheduling system and method for scenic spot | |
| CN112488574B (en) | Travel demand prediction method based on space-time feature extraction | |
| Wang | Taxi scheduling optimization with incomplete information | |
| CN105117806A (en) | Method for acquiring information of remaining tickets of high-speed train and net get-on/get-off passenger flow | |
| Wang et al. | Deep learning for bus passenger demand prediction using big data | |
| CN108959596A (en) | A kind of public transport ladder fare-pricing estimate method |
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 |