CN106408124B - Moving path hybrid prediction method oriented to data sparse environment - Google Patents
Moving path hybrid prediction method oriented to data sparse environment Download PDFInfo
- Publication number
- CN106408124B CN106408124B CN201610841545.4A CN201610841545A CN106408124B CN 106408124 B CN106408124 B CN 106408124B CN 201610841545 A CN201610841545 A CN 201610841545A CN 106408124 B CN106408124 B CN 106408124B
- Authority
- CN
- China
- Prior art keywords
- data
- semantic
- prediction
- predicted
- path
- 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.)
- Expired - Fee Related
Links
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"
Abstract
A moving path hybrid prediction method oriented to a data sparse environment comprises the following steps: acquiring mobile position data information; data processing: data preprocessing and data semantic analysis are carried out on the data; constructing a semantic knowledge base: performing rich semantic conversion and fusion processing on the original trajectory data to construct a semantic knowledge base; constructing a mixed online prediction model: establishing a mixed online prediction model based on forward mode similarity matching calculation and a high-order Markov model based on a semantic knowledge base; and (3) predicted path output: inputting the track segment to be predicted in the mixed online prediction model for prediction, and outputting a prediction path. The method effectively solves the problem of pattern matching failure caused by data sparsity, obviously improves the accuracy of path prediction, and meets the requirements of mobile service application on real-time performance, high efficiency, predictability and the like.
Description
Technical Field
The invention belongs to the technical field of mobile computing, and particularly relates to a moving path hybrid prediction method oriented to a data sparse environment.
Background
At present, with the rapid development and wide popularization of mobile positioning and tracking technology, it is possible to acquire historical trajectory data of a mobile object by using a position sensing device. The value extraction and knowledge discovery of big data of the historical track are realized by semantically modeling and calculating and solving the historical track data, so that the relevant mobile service application is supported, the big data become a remarkable trend and inevitable characteristics in the field of mobile computing, and meanwhile, the big data cause high attention in academic and industrial fields.
Based on large-scale group perception historical track data, the universal and regular movement characteristics and behavior patterns are efficiently mined and extracted, and a movement behavior knowledge base is constructed; by mathematical modeling of the moving track and the maximum probability derivation method, accurate prediction of the online path of the mobile user can be realized. The technology can be widely applied to the fields of urban traffic intelligent scheduling and management, position-based commercial accurate advertisement marketing, tour route recommendation and the like. However, in practical applications, the above method has two problems, which are specifically shown as follows: 1) the difficulty in obtaining context and background knowledge data makes it difficult to improve the prediction accuracy of the movement path by combining a Map Matching (Map Matching technology) or Traffic Flow (Traffic Flow) statistical method; 2) the Data Sparsity Problem (Data Sparsity project) caused by the essential characteristic of low perceived Data value density makes the traditional approach based on approximation calculation or pattern matching difficult to work. The existence of the above problems brings great challenges to application services in the aspect of movement path prediction. Therefore, aiming at the sparsity distribution characteristics of mass perceptual track data, how to construct an accurate prediction model and method of a moving path with strong applicability and high reliability under the condition of missing context and background knowledge data has very important practical significance and application value.
Disclosure of Invention
The invention provides a path prediction method for sparse distribution of mass movement trajectory data, which adopts the technical scheme that:
a moving path hybrid prediction method oriented to a data sparse environment comprises the following steps:
s1: acquiring mobile position data information;
s2: data processing: data preprocessing and data semantic analysis are carried out on the data;
s3: constructing a semantic knowledge base: performing rich semantic conversion and fusion processing on the original trajectory data to construct a semantic knowledge base;
s4: constructing a mixed online prediction model: establishing a mixed online prediction model based on forward mode similarity matching calculation and a high-order Markov model based on a semantic knowledge base;
s5: and (3) predicted path output: inputting the track segment to be predicted in the mixed online prediction model for prediction, and outputting a prediction path.
Further, in a moving path hybrid prediction method under a data sparse environment, the moving position data information in S1 includes a track segment to be predicted.
Further, the moving path hybrid prediction method in the data sparse environment is oriented, wherein the data semantic analysis in S2 includes performing unified semantic coordinate transformation operation on the data, dividing the data into complete moving trajectory segments, and labeling the segments.
Further, a moving path hybrid prediction method for a data sparse environment, where performing rich semantic conversion and fusion processing on original trajectory data in S3 includes: semantic segmentation of a hidden map, extraction of simplified nodes of a road network framework and knowledge discovery of a mobile mode.
Further, a moving path hybrid prediction method oriented to a data sparse environment, wherein the processing of performing rich semantic conversion and fusion on the original trajectory data in S3 further includes an area level Pair migration matrix, and the area level Pair migration matrix is calculated based on implicit map semantic segmentation processing.
Further, a moving path hybrid prediction method oriented to a data sparse environment, where the processing of performing rich semantic conversion and fusion on the original trajectory data in S3 further includes a multiple continuous state transition probability model, where the multiple continuous state transition probability model is used to convert semantic data into a segmented trajectory sequence.
Further, a moving path hybrid prediction method in a data sparse environment is provided, wherein the forward pattern matching in S4 includes element matching and distance calculation.
Further, a moving path hybrid prediction method facing to a data sparse environment preferentially executes a forward mode matching prediction process, and outputs a prediction path when a solution exists in the matching process; and when no solution exists in the matching process, executing a Markov probability inference model, forward recurrently distributing the transition probability of the continuous state of the corresponding order by taking the current moving state as a reference, taking the probability maximum value as an output predicted path with the step length of 1, and outputting the predicted path by taking the predicted destination position information as a termination condition through a recursion cycle process.
Further, in a moving path hybrid prediction method under a data sparse environment, the data processing in S2 is executed before the track segment to be predicted in S5 is input.
Further, a moving path hybrid prediction method oriented to a data sparse environment is designed with respect to data preprocessing and semantic parsing steps in S2, and corresponding dynamic monitoring management and message generation flow control are designed.
The invention provides a path prediction method facing to moving track data sparsity distribution, which is characterized in that a city map semantic model is built through historical track data to realize rich semantic conversion and fusion of original track data, a hybrid prediction method of historical moving Pattern Matching calculation (Pattern Matching calculation) and high Order Markov (high Order Markov model) is built, the essence of the hybrid prediction method is to build a hybrid and replaceable collaborative prediction model, and on one hand, the advantage of forward Pattern similarity Matching in the aspect of time sequence unconstrained historical information is fully utilized to improve the path prediction precision; on the other hand, a segmented track moving state transition probability matrix model under a strict time sequence relation is constructed by a replaceable high-order Markov model, so that the problem of Pattern Matching failure (No Pattern Matching) caused by data sparsity is effectively solved, the accuracy of path prediction is obviously improved, and the requirements of mobile service application on the aspects of instantaneity, high efficiency, predictability and the like are met.
Drawings
FIG. 1 is a schematic diagram illustrating steps of a hybrid prediction method for a moving path in a data sparse environment according to the present invention;
FIG. 2 is a schematic diagram of a forward mode similarity matching result of a moving path hybrid prediction method in a data sparse environment according to the present invention;
FIG. 3 is a path prediction result of a moving path hybrid prediction method under a data sparse environment under different known lengths of on-line prediction tracks;
fig. 4 is a schematic diagram of a prediction result of the moving path hybrid prediction method in the data sparse environment.
Detailed Description
A moving path hybrid prediction method oriented to a data sparse environment comprises the following steps:
s1: acquiring mobile position data information;
the information receives and stores data from multiple source location aware sources (vehicle GPS, mobile smart phone, PDA, etc.). In this embodiment, the movement position data information includes a track segment to be predicted.
S2: data processing: carrying out data preprocessing and data semantic analysis on the data;
as shown in fig. 1: the data is preprocessed as follows: preprocessing the collected historical track data, specifically comprising: 1) the module carries out noise detection and filtering processing due to the signal intensity change of the positioning device and noise data introduced by channel change in the data transmission process; 2) due to the instability of link connection in the process of fast moving of a moving object and the data loss phenomenon caused by the packet loss phenomenon in the data transmission process, the module performs data interpolation operation at equal time intervals;
the data semantic analysis comprises the following steps of carrying out unified semantic coordinate conversion operation on data, dividing the data into complete movement track segments, and labeling: 1) due to the loss of an automatic process and a labeling attribute in the process of acquiring the moving track, the module executes semantic segmentation operation of Complete track segments based on a time interval threshold and a space interval distance threshold on the continuously acquired moving track; 2) due to the difference of the multi-source mobile positioning equipment in the aspect of selecting a space coordinate system (a WGS84 coordinate system, a GGRS87 coordinate system, a GSM coordinate system and the like), the module executes unified semantic coordinate conversion operation.
In this embodiment, corresponding dynamic monitoring management and message generation flow control are designed for the data processing step of step S2, that is, in the process of receiving and analyzing the moving trajectory data, the flow controller controls the flow operation of the data receiving and semantic analyzing process.
S3: constructing a semantic knowledge base: performing rich semantic conversion and fusion processing on the original trajectory data to construct a semantic knowledge base; in S3, performing semantic-rich transformation and fusion processing on the original trajectory data includes:
as shown in fig. 3, the specific implementation process of semantic segmentation of the implicit map includes: calculating a two-dimensional density function based on the spatial distribution characteristics of the large-scale historical offline track position points, and realizing the reconstruction of the geographic spatial semantic topological relation of the implicit map and the secondary division process of the area level spatial area by combining the two-dimensional density function and the boundary approximation value set;
extracting simplified nodes of a road network framework: similar to a hidden map semantic segmentation module, hidden position points of movement-limited track data under the constraint of a road network are extracted based on the spatial distribution density characteristics of large-scale movement track data to form a road network skeleton simple node set.
In the road network skeleton simple node extraction process, key road network nodes are extracted through a density-based clustering method on the basis of calculation of moving corners of adjacent k sequences in a historical moving track.
The moving pattern knowledge discovery specifically comprises the steps of carrying out rich semantic conversion process on the originally acquired moving track data on the basis of hidden map semantic segmentation and road network skeleton simple node output set, further efficiently extracting a potential moving behavior pattern set on the basis of a sequence pattern mining algorithm, and providing knowledge support for subsequent moving path prediction.
The method comprises the steps of performing rich semantic conversion and fusion processing on original trajectory data in S3, and further comprising an area-level Pair migration matrix, wherein the area-level Pair migration matrix is obtained by calculation based on implicit map semantic segmentation processing, and is specifically realized by taking a Complete trajectory segment set in a mobile trajectory data semantic analysis module as input data, extracting a Pair point set, further realizing a rich semantic conversion process of a corresponding Pair point set based on the area division of the area-level space of the implicit map semantic, and constructing a corresponding Pair migration probability matrix block under the conditional probability by using a Bayesian network. Where pair refers to a pair from one location to another, containing the start and end points of the response; a pair at a zone level refers to starting from one zone and ending at another zone.
The processing of rich semantic conversion and fusion on the original trajectory data in S3 further includes a multiple continuous state transition probability model, where the multiple continuous state transition probability model is used to convert the segmented trajectory sequence into semantic data.
The semantic knowledge base is high-order state motion transition matrix knowledge which is output by a motion behavior mode set, a motion behavior association rule and multiple continuous state transition probabilities and is based on a Markov probability process, wherein the motion behavior mode set, the motion behavior association rule and the multiple continuous state transition probabilities are discovered by the motion mode knowledge discovery module and stored.
S4: constructing a mixed online prediction model: establishing a mixed online prediction model based on forward mode similarity matching calculation and a high-order Markov model based on a semantic knowledge base;
the forward mode matching comprises element matching and distance calculation, is used for realizing a similarity calculation process between the Partial track to be predicted and the found moving mode based on the matching direction and the distance, and simultaneously outputs a moving path prediction result through comparison of a similarity threshold value.
Wherein the Markov probability prediction model: the method is used for realizing the probability calculation and prediction process between the Partial track to be predicted and the counted multi-continuous-state transition probability matrix by taking the current moving state as a reference and taking the high-order Markov order as a step length, and simultaneously outputting a moving path prediction result with the step length of 1.
In the embodiment, a forward mode matching prediction process is preferentially executed, historical movement information in Partial is utilized to the maximum extent, so that the precision of the output prediction path is effectively improved, wherein the forward mode matching process compares the similarity degree between the movement pattern and the Partial fragment one by one through the matching of the same element between the Partial track fragment and the found movement pattern and the forward distance calculation of the same matching element, and corresponding suffix sequences are returned as the output prediction path under the condition that the matching process has a solution; and under the condition that no solution exists in the matching process, executing a Markov probability inference model, forward recurrently distributing the transition probability distribution of the continuous state of the corresponding order by taking the current moving state as a reference, taking the probability maximum value as the output predicted path with the step length of 1, and generating the final output predicted path by taking the deduced Partial destination position information as a termination condition through a recursive circulation process.
In the hybrid prediction model, the forward mode similarity matching calculation method is as follows:
in the above formula (1), degree represents a similarity value between the historical movement pattern and the online segment movement track, cov represents a pattern matching length of the historical movement pattern and the online segment movement track, dis represents a composite distance between the current position of the online query track and the historical pattern, and sup represents a support degree of the historical pattern. In the formula (2), ekFor all elements in the history schema that match the online query trajectory, eendRepresenting the current location of the online query trajectory.
Briefly explaining the foregoing forward pattern similarity matching process, as shown in fig. 2, a circular sequence is an online query movement trajectory segment, and there are 3 matchable patterns in the historical movement patterns, respectively, where a triangle represents a matching element, and a square represents a short-term prediction path, where a value of a candidate pattern 1 is 2 and is 1.
In a high-order Markov calculation model, a next potential position Rank calculation formula with the step length of 1 is as follows:
arg Max:score(loc)
wherein score (loc) represents the Rank value of the candidate location loc, dorigRepresents the distance of the position loc from the start position of the segment trajectory, ddestPro (loc) is the migration probability value of location loc in the trained high-order Markov model. And calculating the Rank values of the m candidate positions loc, and taking the position with the largest Rank value as the next predicted position of the online query fragment track.
S5: and (3) predicted path output: inputting the track segment to be predicted in the mixed online prediction model for prediction, and outputting a prediction path.
Aiming at the sparsity distribution characteristics of the movement track data, the purpose of predicting the future movement path of the online inquiry track segment is realized by the mixed and complementary prediction mode. The method is characterized in that the comparison and verification of the predicted path results are respectively carried out on 5000 test track segments under the five conditions that the known lengths are 10%, 20%, 30%, 40% and 50%, and the results are respectively compared with a 1-order Markov model method and a 2-order Markov model method, so that the Hybrid Moving Route Prediction (HMRP) constructed by the method has remarkable advantages, and is shown in fig. 3 and 4.
Claims (8)
1. A moving path hybrid prediction method oriented to a data sparse environment is characterized by comprising the following steps: the method comprises the following steps:
s1: acquiring mobile position data information;
s2: data processing: data preprocessing and data semantic analysis are carried out on the data;
s3: constructing a semantic knowledge base: performing rich semantic conversion and fusion processing on the original trajectory data to construct a semantic knowledge base;
s4: constructing a mixed online prediction model: establishing a mixed online prediction model based on forward mode similarity matching and a high-order Markov model based on a semantic knowledge base;
preferentially executing a forward mode similarity matching prediction process, and outputting a prediction path when a solution exists in the matching process; when no solution exists in the matching process, a high-order Markov model is executed, the current moving state is taken as a reference, the transition probability distribution of the continuous state of the corresponding order is forward recurred, the probability maximum value is taken as a predicted path with the output step length being 1, and the predicted path is output by taking the predicted destination position information as a termination condition through a recursion cycle process;
the forward pattern similarity matching comprises element matching and distance calculation;
in the mixed online prediction model, the forward mode similarity matching calculation method is as follows:
in the above formula, degree represents the similarity value between the historical movement pattern and the online segment movement track, dis represents the composite distance between the current position of the online query track and the historical pattern, and sup is the support degree of the historical pattern; e.g. of the typekFor all elements in the history schema that match the online query trajectory, eendCurrent position of online query track is shown as cov ekAnd eendThe pattern matching length of the two;
in the high-order markov model, the next potential position Rank calculation formula with the step size of 1 is as follows:
arg Max:score(loc)
wherein score (loc) represents the Rank value of the candidate location loc, dorigRepresents the distance of the position loc from the start position of the segment trajectory, ddestDeducing a distance of the destination for the location loc from the trajectory of the segment, pro (loc) being a probability value of the migration of the location loc in the trained high order Markov model; calculating the Rank values of the m candidate positions loc, and taking the position with the largest Rank value as the next predicted position of the online query fragment track;
s5: and (3) predicted path output: inputting the track segment to be predicted in the mixed online prediction model for prediction, and outputting a prediction path.
2. The moving path hybrid prediction method oriented to the data sparse environment according to claim 1, wherein: the moving position data information in S1 includes the track segment to be predicted.
3. The moving path hybrid prediction method oriented to the data sparse environment according to claim 1, wherein: and the data semantic analysis in the step S2 includes performing unified semantic coordinate conversion operation on the data, dividing the data into complete movement track segments, and labeling the segments.
4. The moving path hybrid prediction method oriented to the data sparse environment according to claim 1, wherein: in S3, performing semantic-rich transformation and fusion processing on the original trajectory data includes: semantic segmentation of a hidden map, extraction of simplified nodes of a road network framework and knowledge discovery of a mobile mode.
5. The moving path hybrid prediction method oriented to the data sparse environment according to claim 4, wherein: the rich semantic conversion and fusion processing of the original trajectory data in the step S3 further includes an area-level Pair migration matrix, and the area-level Pair migration matrix is obtained by performing semantic segmentation processing and calculation based on an implicit map.
6. The moving path hybrid prediction method oriented to the data sparse environment according to claim 4, wherein: the processing of rich semantic conversion and fusion on the original trajectory data in S3 further includes a multiple continuous state transition probability model, where the multiple continuous state transition probability model is used to convert the segmented trajectory sequence into semantic data.
7. The moving path hybrid prediction method oriented to the data sparse environment according to claim 1, wherein: and executing the corresponding data processing step in S2 before the track segment to be predicted in the S5 is input.
8. The moving path hybrid prediction method oriented to the data sparse environment according to any one of claims 1 to 7, characterized in that: corresponding dynamic monitoring management and message generation flow control are designed aiming at the steps of data preprocessing and semantic parsing in S2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610841545.4A CN106408124B (en) | 2016-09-22 | 2016-09-22 | Moving path hybrid prediction method oriented to data sparse environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610841545.4A CN106408124B (en) | 2016-09-22 | 2016-09-22 | Moving path hybrid prediction method oriented to data sparse environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106408124A CN106408124A (en) | 2017-02-15 |
| CN106408124B true CN106408124B (en) | 2020-01-10 |
Family
ID=57996884
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610841545.4A Expired - Fee Related CN106408124B (en) | 2016-09-22 | 2016-09-22 | Moving path hybrid prediction method oriented to data sparse environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106408124B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019084918A1 (en) | 2017-11-03 | 2019-05-09 | Beijing Didi Infinity Technology And Development Co., Ltd. | System and method for determining a trajectory |
| CN108981739B (en) * | 2018-06-08 | 2022-02-22 | 南方科技大学 | Path planning method, device, server and storage medium |
| CN109034448B (en) * | 2018-06-14 | 2022-02-11 | 重庆邮电大学 | Trajectory prediction method based on vehicle trajectory semantic analysis and deep belief network |
| CN109190656B (en) * | 2018-07-16 | 2020-07-21 | 浙江大学 | Indoor semantic track marking and complementing method under low-sampling positioning environment |
| CN109697221B (en) * | 2018-11-22 | 2021-07-09 | 东软集团股份有限公司 | Track law mining method and device, storage medium and electronic equipment |
| CN109714263B (en) * | 2019-01-18 | 2021-01-29 | 北京邮电大学 | Path selection method and device in satellite communication network |
| CN110321343B (en) * | 2019-07-11 | 2023-11-14 | 广东工业大学 | Wearable device track prediction method, device and system |
| CN110475208B (en) * | 2019-08-06 | 2021-07-16 | 上海尚往网络科技有限公司 | Method and device for determining travel track of user equipment |
| CN110909592B (en) * | 2019-10-11 | 2020-12-18 | 重庆特斯联智慧科技股份有限公司 | Target tracking method and system based on multi-scale characteristic quantity |
| CN110909913B (en) * | 2019-10-11 | 2021-06-04 | 重庆特斯联智慧科技股份有限公司 | Scenic spot tour guide service pre-estimation system and method based on intelligent tracking |
| CN110795519A (en) * | 2019-10-28 | 2020-02-14 | 天聚地合(苏州)数据股份有限公司 | Markov model and probability statistics-based position prediction method and readable storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102713520A (en) * | 2009-10-12 | 2012-10-03 | 通腾科技股份有限公司 | Navigation device & method |
| CN103177562A (en) * | 2011-12-26 | 2013-06-26 | 中国移动通信集团公司 | Method and device for obtaining information of traffic condition prediction |
| CN103415070A (en) * | 2013-07-04 | 2013-11-27 | 百度在线网络技术（北京）有限公司 | Positioning method and device based on mobile path |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9049134B2 (en) * | 2013-03-08 | 2015-06-02 | Disney Enterprises, Inc. | Network condition predictions for multimedia streaming |
-
2016
- 2016-09-22 CN CN201610841545.4A patent/CN106408124B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102713520A (en) * | 2009-10-12 | 2012-10-03 | 通腾科技股份有限公司 | Navigation device & method |
| CN103177562A (en) * | 2011-12-26 | 2013-06-26 | 中国移动通信集团公司 | Method and device for obtaining information of traffic condition prediction |
| CN103415070A (en) * | 2013-07-04 | 2013-11-27 | 百度在线网络技术（北京）有限公司 | Positioning method and device based on mobile path |
Non-Patent Citations (2)
| Title |
|---|
| 用于移动路径预测的混合Markov模型;余雪岗等;《通信学报》;20061231;第27卷(第12期);第61-68页 * |
| 融合复合特征的移动轨迹预测方法的研究与实现;张盼盼;《万方数据》;20160504;第1-49页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106408124A (en) | 2017-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106408124B (en) | Moving path hybrid prediction method oriented to data sparse environment | |
| Wang et al. | When will you arrive? estimating travel time based on deep neural networks | |
| Zhao et al. | Predictive task assignment in spatial crowdsourcing: a data-driven approach | |
| Liu et al. | DeepVM: RNN-based vehicle mobility prediction to support intelligent vehicle applications | |
| Ding et al. | Network-matched trajectory-based moving-object database: Models and applications | |
| Zhu et al. | Inferring taxi status using gps trajectories | |
| CN114428828A (en) | Method and device for digging new road based on driving track and electronic equipment | |
| CN113159403B (en) | Intersection pedestrian track prediction method and device | |
| CN113068131B (en) | Method, device, equipment and storage medium for predicting user movement mode and track | |
| Wang et al. | Vehicle trajectory prediction algorithm in vehicular network | |
| CN110929951A (en) | Correlation analysis and prediction method for power grid alarm signal | |
| Huang et al. | Multimodal trajectory prediction: A survey | |
| Bansal et al. | Htfm: Hybrid traffic-flow forecasting model for intelligent vehicular ad hoc networks | |
| CN109581444B (en) | GPS track segmentation and semantic annotation method | |
| Singh et al. | Multi-scale graph-transformer network for trajectory prediction of the autonomous vehicles | |
| Ban et al. | A deep concept graph network for interaction-aware trajectory prediction | |
| Xu et al. | MVHGN: Multi-view adaptive hierarchical spatial graph convolution network based trajectory prediction for heterogeneous traffic-agents | |
| Zhou et al. | CSR: cascade conditional variational auto encoder with socially-aware regression for pedestrian trajectory prediction | |
| Chen et al. | Deep learning for trajectory data management and mining: A survey and beyond | |
| Sharma et al. | Nearest neighbour classification for trajectory data | |
| Chen et al. | Digital twin mobility profiling: A spatio-temporal graph learning approach | |
| Hu et al. | Recurrent multi-graph neural networks for travel cost prediction | |
| Yang et al. | Real‐time detection of traffic congestion based on trajectory data | |
| Yan et al. | A Survey of Generative AI for Intelligent Transportation Systems | |
| Dutta et al. | CLUSTMOSA: Clustering for GPS trajectory data based on multi-objective simulated annealing to develop mobility application |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200110 Termination date: 20210922 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |