CN108108831A - A kind of destination Forecasting Methodology and device - Google Patents
A kind of destination Forecasting Methodology and device Download PDFInfo
- Publication number
- CN108108831A CN108108831A CN201611050832.XA CN201611050832A CN108108831A CN 108108831 A CN108108831 A CN 108108831A CN 201611050832 A CN201611050832 A CN 201611050832A CN 108108831 A CN108108831 A CN 108108831A
- Authority
- CN
- China
- Prior art keywords
- historical position
- position point
- tensor
- network
- dimension
- 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
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"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention relates to intelligent transportation fields more particularly to a kind of destination Forecasting Methodology and device, this method to be:Pre-set destination prediction model,Wherein,The goal displacement probability between each two historical position point in each history wheelpath of vehicle in each set period of time is had recorded in above destination prediction model respectively,By the current driving track for monitoring vehicle,Determine the current driving track corresponding period,And based on the above-mentioned current driving track corresponding period,Using default destination prediction model,The prediction probability between each historical position point in above-mentioned current driving track and each history wheelpath is calculated respectively,The historical position point that corresponding prediction probability is met to preset condition is determined as predicting destination,So,When whether establishing destination prediction model,Or when carrying out destination prediction,All take into account time factor,Effectively improve the accuracy of destination prediction,And then improve the driving experience of user.
Description
Technical field
The present invention relates to intelligent transportation field more particularly to a kind of destination Forecasting Methodology and devices.
Background technology
With the development of Internet technology, the application journey based on location-based service (Location Based Service, LBS)
Sequence is constantly emerged in large numbers, and the clothes such as position positioning, route inquiry and the presentation of history wheelpath are provided so as to drive to go on a journey for user
Business.
And since the drive route of trip of user can both be subject to the custom of itself behavior to influence, also suffer from external condition
Constraint, therefore, user drive trip destination be that there is certain rule, user drives to reach some specific regions
Possibility is higher, e.g., family, company, shopping center, dining room and cinema etc..
Under the prior art, terminal can by the application program of location-based service obtain user drive trip history drive a vehicle rail
Mark, it is possible to further based on user drive trip history wheelpath prediction user drive trip destination.
Under the prior art, predictably terminal imagination user drive trip destination mode it is as follows:
Terminal obtains the current wheelpath of vehicle and each history wheelpath, and by current wheelpath with
Each history wheelpath is compared to pair, if there are a certain history wheelpaths and current wheelpath part to compare success,
The destination for then determining current vehicle is the destination of the history wheelpath.Further, if there are multiple history wheelpaths
Success is compared with current wheelpath part, then according to above-mentioned multiple history wheelpaths, it is above-mentioned multiple to calculate arrival correspondence
The probability of each destination of history wheelpath determines the destination of current vehicle for probability supreme good.
But due to user drive trip track be accustomed to by user, purpose and demand etc. are affected, Er Qie
When carrying out track comparison, there is no the otherness considered by way of ground and destination, and in fact, purpose in history wheelpath
Ground more can be shown that user currently drives the destination gone on a journey, also not accounting in history wheelpath for by way of ground
Time factor, and in fact, the history wheelpath that on the one hand different time sections of one day generate can reflect that the trip of user is practised
On the other hand used and different traffic, can more reflect that user is current closer to the history wheelpath of current period and drive
Sail custom, therefore, prediction user drive trip destination accuracy rate than relatively low.Further, due to the current line of vehicle
Wheel paths are continually changing, therefore, are compared when can not find in the database in history wheelpath with current wheelpath
During successful history wheelpath, then unpredictable user drives the destination of trip.
In view of this, it is necessary to a kind of new destination Forecasting Methodology be designed, to overcome drawbacks described above.
The content of the invention
The embodiment of the present invention provides a kind of destination Forecasting Methodology and device, for the current of the vehicle that is driven according to user
Wheelpath and history wheelpath, prediction user drive the destination gone on a journey, to solve and drive a vehicle by history in the prior art
Track carry out destination prediction when, destination prediction accuracy it is too low the problem of.
Specific technical solution provided in an embodiment of the present invention is as follows:
A kind of destination Forecasting Methodology, including:
The current driving track of vehicle is monitored, determines the current driving track corresponding period;
Default destination prediction model is obtained, each setting time is had recorded respectively in the destination prediction model
In section, in each history wheelpath of the vehicle, the goal displacement probability between each two historical position point, wherein, institute
Stating set period of time was divided based on a consecutive days;
Based on the current driving track corresponding period, using the destination prediction model, calculate respectively described in
Prediction probability in current driving track and each history wheelpath between each historical position point;
The historical position point that corresponding prediction probability is met to preset condition is determined as predicting destination.
Optionally, before the current driving track for monitoring the vehicle, further comprise:Generate the destination prediction mould
Type specifically includes:
Obtain each history wheelpath of the vehicle;
Each history wheelpath of the vehicle of acquisition is projected to road network, obtains goal displacement network,
Wherein, the goal displacement network is used to indicate opposite between each historical position point in each history wheelpath
Movement;
First transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, first dimension represents
Each historical position point before being shifted in the goal displacement network, second dimension represent the goal displacement network
It is middle shift after each historical position point, third dimension represents set period of time set, and first transport tensor is used for
Indicate in each set period of time, in each history wheelpath of the vehicle between each two historical position point first
Transition probability;
First transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, wherein, the mesh
Mark transport tensor is had recorded respectively in each set period of time, and in each history wheelpath of the vehicle, each two is gone through
Goal displacement probability between history location point.
Optionally, each history wheelpath of the vehicle of acquisition is projected to road network, obtains target and turn
Network is moved, including:
Each history wheelpath of the vehicle of acquisition is projected to road network, road network in the projected
In establish the first transfer network;
Each historical position point in each history wheelpath included in network is shifted by described first to identify
For transient state historical position point, and the absorption historical position point of a mirror image is created for each transient state historical position point, wherein, institute
The nonterminal point in transient state historical position point expression history wheelpath is stated, the absorption historical position point represents history driving rail
Terminating point in mark;
Based between each transient state historical position point in each history wheelpath and each absorption historical position point
Relative motion, establish corresponding goal displacement network.
Optionally, the first transport tensor is established based on the first dimension, the second dimension and third dimension, including:
By the goal displacement network according to default time cut-point, long-term goal transfer network and recent mesh are divided into
Mark transfer network, wherein, the long-term goal transfer network includes in the goal displacement network in the time cut-point
The historical position point generated before, the immediate objective transfer network included in the goal displacement network in the time point
The historical position point generated after cutpoint;
Network is shifted according to the long-term goal, and at a specified future date the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of first transport tensor at a specified future date represents to occur in the long-term goal transfer network
Each historical position point before transfer, the second dimension of first transport tensor at a specified future date represent the long-term goal transfer network
It is middle shift after each historical position point, the third dimension of first transport tensor at a specified future date represents the set period of time
Set, first transport tensor at a specified future date are used to indicate in each set period of time before the time cut-point, institute
State the first transition probability of long term between each two historical position point in each history wheelpath of vehicle;
Network is shifted according to the immediate objective, and recent the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of recent first transport tensor represents to occur in the immediate objective transfer network
Each historical position point before transfer, the second dimension of recent first transport tensor represent the immediate objective transfer network
It is middle shift after each historical position point, the third dimension of recent first transport tensor represents the set period of time
Set, recent first transport tensor are used to indicate in each set period of time after the time cut-point, institute
State recent first transition probability between each two historical position point in each history wheelpath of vehicle;
Using first transport tensor at a specified future date and recent first transport tensor, first transport tensor is formed.
Optionally, first transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, bag
It includes:
The vehicle fleet size that occurs in the goal displacement network in each set period of time is obtained respectively and described
The specified point of interest quantity being located in goal displacement network in each historical position point preset range forms central tensor;
The central tensor based on acquisition, is filled first transport tensor, obtains the second transport tensor;
Training is optimized to second transport tensor of acquisition, obtains each optimization training result;
Based on each optimization training result, the optimization training result of preset rules is determined for compliance with as goal displacement
Amount.
Optionally, training is optimized to second transport tensor of acquisition, obtains each optimization training result, wrapped
It includes:
Calculate the penalty values between first transport tensor and second transport tensor;
Based on the penalty values and each regularization term, second transport tensor is fitted, obtains each optimization
Training result, wherein, each regularization term is by the set period of time set, each history of vehicle driving rail
Each historical position point, the goal displacement network, the central default geographical feature factor of tensor sum generate in mark.
Optionally, based on the current driving track corresponding period, using the destination prediction model, institute is calculated
The prediction probability between any one historical position point in current driving track and each history wheelpath is stated, including:
Determine that source location set, current stop place point and the source location set of the current driving track are stopped with current
Each between location point passes by location point;
It determines in the destination prediction model, within the corresponding period of the current driving track, the current line
The source location set of wheel paths, current stop place point and it is each pass by location point, each with each history wheelpath
Described in goal displacement probability between any one historical position point;
Each goal displacement probability based on acquisition determines the current driving track and each history driving rail
Prediction probability described in mark between any one historical position point.
A kind of destination prediction meanss, including:
Monitoring unit for monitoring the current driving track of vehicle, determines the current driving track corresponding period;
Acquiring unit for obtaining default destination prediction model, has recorded respectively in the destination prediction model
In each set period of time, in each history wheelpath of the vehicle, the target between each two historical position point turns
Probability is moved, wherein, the set period of time was divided based on a consecutive days;
Computing unit, for being based on the current driving track corresponding period, using the destination prediction model,
The prediction probability between each historical position point in the current driving track and each history wheelpath is calculated respectively;
Determination unit, the historical position point for corresponding prediction probability to be met to preset condition are determined as predicting purpose
Ground.
Optionally, described device further includes:Model foundation unit;
The model foundation unit is used for, and before the current driving track of the vehicle is monitored, generates the destination
Prediction model specifically includes:
Obtain each history wheelpath of the vehicle;
Each history wheelpath of the vehicle of acquisition is projected to road network, obtains goal displacement network,
Wherein, the goal displacement network is used to indicate opposite between each historical position point in each history wheelpath
Movement;
First transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, first dimension represents
Each historical position point before being shifted in the goal displacement network, second dimension represent the goal displacement network
It is middle shift after each historical position point, third dimension represents set period of time set, and first transport tensor is used for
Indicate in each set period of time, in each history wheelpath of the vehicle between each two historical position point first
Transition probability;
First transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, wherein, the mesh
Mark transport tensor is had recorded respectively in each set period of time, and in each history wheelpath of the vehicle, each two is gone through
Goal displacement probability between history location point.
Optionally, each history wheelpath of the vehicle of acquisition is projected to road network, obtains target and turn
When moving network, the model foundation unit is used for:
Each history wheelpath of the vehicle of acquisition is projected to road network, road network in the projected
In establish the first transfer network;
Each historical position point in each history wheelpath included in network is shifted by described first to identify
For transient state historical position point, and the absorption historical position point of a mirror image is created for each transient state historical position point, wherein, institute
The nonterminal point in transient state historical position point expression history wheelpath is stated, the absorption historical position point represents history driving rail
Terminating point in mark;
Based between each transient state historical position point in each history wheelpath and each absorption historical position point
Relative motion, establish corresponding goal displacement network.
Optionally, when establishing the first transport tensor based on the first dimension, the second dimension and third dimension, the model foundation
Unit is used for:
By the goal displacement network according to default time cut-point, long-term goal transfer network and recent mesh are divided into
Mark transfer network, wherein, the long-term goal transfer network includes in the goal displacement network in the time cut-point
The historical position point generated before, the immediate objective transfer network included in the goal displacement network in the time point
The historical position point generated after cutpoint;
Network is shifted according to the long-term goal, and at a specified future date the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of first transport tensor at a specified future date represents to occur in the long-term goal transfer network
Each historical position point before transfer, the second dimension of first transport tensor at a specified future date represent the long-term goal transfer network
It is middle shift after each historical position point, the third dimension of first transport tensor at a specified future date represents the set period of time
Set, first transport tensor at a specified future date are used to indicate in each set period of time before the time cut-point, institute
State the first transition probability of long term between each two historical position point in each history wheelpath of vehicle;
Network is shifted according to the immediate objective, and recent the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of recent first transport tensor represents to occur in the immediate objective transfer network
Each historical position point before transfer, the second dimension of recent first transport tensor represent the immediate objective transfer network
It is middle shift after each historical position point, the third dimension of recent first transport tensor represents the set period of time
Set, recent first transport tensor are used to indicate in each set period of time after the time cut-point, institute
State recent first transition probability between each two historical position point in each history wheelpath of vehicle;
Using first transport tensor at a specified future date and recent first transport tensor, first transport tensor is formed.
Optionally, first transport tensor is handled, when acquisition meets the goal displacement tensor of preset rules, institute
Model foundation unit is stated to be used for:
The vehicle fleet size that occurs in the goal displacement network in each set period of time is obtained respectively and described
The specified point of interest quantity being located in goal displacement network in each historical position point preset range forms central tensor;
The central tensor based on acquisition, is filled first transport tensor, obtains the second transport tensor;
Training is optimized to second transport tensor of acquisition, obtains each optimization training result;
Based on each optimization training result, the optimization training result of preset rules is determined for compliance with as goal displacement
Amount.
Optionally, training is optimized to second transport tensor of acquisition, when obtaining each optimization training result, institute
Model foundation unit is stated to be used for:
Calculate the penalty values between first transport tensor and second transport tensor;
Based on the penalty values and each regularization term, second transport tensor is fitted, obtains each optimization
Training result, wherein, each regularization term is by the set period of time set, each history of vehicle driving rail
Each historical position point, the goal displacement network, the central default geographical feature factor of tensor sum generate in mark.
Optionally, based on the current driving track corresponding period, using the destination prediction model, institute is calculated
When stating in current driving track and each history wheelpath the prediction probability between any one historical position point, the calculating
Unit is used for:
Determine that source location set, current stop place point and the source location set of the current driving track are stopped with current
Each between location point passes by location point;
It determines in the destination prediction model, within the corresponding period of the current driving track, the current line
The source location set of wheel paths, current stop place point and it is each pass by location point, each with each history wheelpath
Described in goal displacement probability between any one historical position point;
Each goal displacement probability based on acquisition determines the current driving track and each history driving rail
Prediction probability described in mark between any one historical position point.
The present invention has the beneficial effect that:
In the embodiment of the present invention, each history wheelpath of vehicle and above-mentioned each history driving rail are in advance based on
The otherness of the time factor of mark, terminating point and nonterminal point, establishes a destination prediction model, wherein, above destination
It is had recorded respectively in each set period of time in prediction model, in each history wheelpath of the vehicle, each two is gone through
Goal displacement probability between history location point when monitoring the current driving track of vehicle, determines above-mentioned current driving track pair
The period answered, and based on above-mentioned current driving track and above-mentioned current track corresponding period, using pre-set mesh
Ground prediction model, calculate respectively in above-mentioned current driving track and each history wheelpath between each historical position point
Prediction probability, by corresponding prediction probability meet preset condition historical position point be determined as predict destination, in this way, can examine
Consider the otherness of nonterminal point and terminating point in historical behavior track, at the same can also fully take into account historical behavior track when
Between factor so that the destination prediction model of foundation is more accurate, so as to improve destination prediction accuracy, and then promoted
The driving experience of user.
Description of the drawings
Fig. 1 is the flow chart that destination prediction model is established in the embodiment of the present invention;
Fig. 2 is the schematic diagram of the first transport tensor in the embodiment of the present invention;
Fig. 3 is the flow chart for the method that destination is predicted in the embodiment of the present invention;
Fig. 4 is the structure diagram for the device that destination is predicted in the embodiment of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, is not whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work
Embodiment belongs to the scope of protection of the invention.
In order to improve the accuracy of destination prediction, in the embodiment of the present invention, a kind of method for predicting destination is devised,
This method is first to preset destination prediction model, wherein, when each setting is had recorded in above destination prediction model respectively
Between in section, in each history wheelpath of vehicle, the goal displacement probability between each two historical position point passes through monitoring car
Current driving track, determine the current driving track corresponding period, and based on above-mentioned current driving track it is corresponding when
Between section, using default destination prediction model, calculate respectively above-mentioned current driving track with it is every in each history wheelpath
Prediction probability between one historical position point, by corresponding prediction probability meet preset condition historical position point be determined as it is pre-
Survey destination.
In the embodiment of the present invention, the vehicle driven using terminal-pair user is monitored, and records the driving of vehicle at any time
Track, through accumulation after a while, several history wheelpaths can be recorded in terminal, based on these history driving rail
Mark, terminal can establish destination prediction model, in this way, during subsequent vehicle traveling, terminal just can be in time according to vehicle
Current driving trajectory predictions in front of destination, wherein, above-mentioned terminal can be car-mounted terminal or intelligent hand
Machine, Intelligent bracelet, tablet computer etc..
Specifically, as shown in fig.1, the terminal detailed process of establishing above destination prediction model is as follows:
Step 100:Terminal obtains the history wheelpath for the vehicle that user drives, and obtained each history is driven a vehicle
Track is projected in road network.
Specifically, terminal is to obtain user by global positioning system (Global Positioning System, GPS)
Each history wheelpath of the vehicle of driving, what it is due to GPS offers is all that some include longitude, latitude etc. information
Number of coordinates strong point, therefore, each history wheelpath that terminal obtains all are made of a series of number of coordinates strong points, Wu Fajing
Really intuitively reflect the specific travel route of vehicle.
Further, each history wheelpath that terminal will obtain, according in above-mentioned each history wheelpath
Comprising number of coordinates strong point in corresponding coordinate, project in road network, obtain vehicle each history driving rail
Specific travel route of the mark in road network.
In the embodiment of the present invention, preferred embodiment is using map-matching algorithm (e.g., semidefiniteness method, probability system
Calculating method, algorithm for pattern recognition etc.) it is projected, but in the specific implementation, it can be without being limited thereto.
Step 101:The first transfer network is established in the road network of terminal in the projected, wherein, above-mentioned first transfer net
Network is used to indicate, the relative motion between each historical position point before not converting in above-mentioned each history wheelpath.
Specifically, particular row of the terminal according to each history wheelpath of the above-mentioned vehicle of acquisition in road network
Sail route, by the specific travel route of vehicle by way of any one road junction (e.g., four crossway, T-shaped road junction) determine
For a historical position point, the continuous path for connecting any two historical position point is determined as a transfer side, all history
Transfer side between location point and all any two historical position points constitutes the first transfer network, above-mentioned first transfer network
It is used to indicate, the relative motion between each historical position point before not converting in each history wheelpath.
In the embodiment of the present invention, the representation of preferred first transfer network is denoted as G ' (V ', E ') for digraph,
In, V ' represents one group of historical position point, and E ' represents the transfer line set of connection any two historical position point, and, all transfers
While being vector, there is direction.
For example, it is assumed that V ' contains 5 historical position points, it is respectively v1、v2、v3、v4、v5If v1To v3Between there are one
A continuous path e, and v3To v4Between there are a continuous path f, then, E ' contain 2 transfer sides, be respectively e and f.
Step 102:First transfer network of foundation is converted into goal displacement network by terminal, wherein, above-mentioned goal displacement
Network is used to indicate, the relative motion between each historical position point after conversion in above-mentioned each history wheelpath.
Specifically, not to historical position in the one group of historical position point included due to the first transfer network of terminal foundation
The attribute of point distinguishes, i.e. and it is terminating point or nonterminal point not distinguish historical position point, and in fact, carrying out mesh
Ground prediction when, for terminating point compared with nonterminal point more for referential, therefore, it is necessary to the attributes to historical position point to carry out area
Point, when carrying out destination prediction so that guarantee is follow-up, make full use of the otherness of terminating point and nonterminal point so that prediction result is more
Accurately.
In the embodiment of the present invention, preferred embodiment is, using the absorption shape in mirror sink Markov chain model
State specific type distinguishes the one group of historical position point included in the above-mentioned first transfer network, due in mirror sink horse
In the absorbing state specific type of Er Kefu chain models, the node that can not be lost is known as to absorb node, non-absorbing node is known as
Transient state node, and in fact, terminating point is to belong to lose a little, therefore, the historical position point that will be indicated as terminating point marks
To absorb historical position point, the historical position point of nonterminal point is will be indicated as labeled as transient state historical position point, correspondingly, above-mentioned
The transfer line set of first transfer network also follows the conversion of historical position point, contains all connect between transient state historical position points
Transfer while all connection transient state historical position points to when absorbing the transfer between historical position point.
However, the dividing into for each historical position point one-way of the above-mentioned first transfer network is absorbed into historical position point
Or transient state historical position point, it can not avoid, there are a transfering node in a certain history wheelpath, it is a certain being not only this
The nonterminal point of history wheelpath, while be also the situation of the terminating point of an other history wheelpath, i.e. each
Historical position point all may be terminating point, it is also possible to nonterminal point, therefore, and in the embodiment of the present invention, preferred embodiment
For using the mirror image concept in mirror sink Markov chain model, for each history bit in the above-mentioned first transfer network
It puts and a little creates corresponding mirror nodes, wherein, above-mentioned image node is used to distinguish the absorbing state and transient state of historical position point
State, i.e. any one historical position point of attribute differentiation will not be made originally, divide into a transient state historical position point and one
Absorb historical position point, based on each transient state historical position point in each history wheelpath and each absorption historical position point it
Between relative motion, establish corresponding goal displacement network, correspondingly, in the above-mentioned goal displacement network of foundation shift side collection
It closes and contains transfer sides between all connection transient state historical position points and all connection transient state historical position points to absorbing history bit
Transfer side between putting a little.
For example, it is assumed that the first transfer network representation is digraph G ' (V ', E '), wherein, V is expressed as n historical position point,
E is expressed as m transfer side, if goal displacement network representation is digraph G (V, E), then, according to transformational relation, V can there are n
A transient state historical position point and n absorption historical position point, if continuing to assume only has a history in above-mentioned goal displacement network
Wheelpath, and an above-mentioned history wheelpath is by v0To vn-1Common n transient state historical position point composition, then, it is respectively v0
To vn-1Common n transient state historical position point creates corresponding mirror nodes as historical position point is absorbed, such as, it is assumed that transient state
Historical position point viIt is v in goal displacement network for i-th of transient state historical position point in n transient state historical position pointiIt creates
One mirror nodes v2i-1, and so on understand, v0To vn-1The mirror nodes of common n transient state historical position point are respectively vnExtremely
v2n-1。
So far, the first transfer network of foundation is converted to goal displacement network by terminal.
Step 103:Terminal establishes first according to goal displacement network based on the first dimension, the second dimension and third dimension
Transport tensor, wherein, above-mentioned first transport tensor is used to indicate each history row of above-mentioned vehicle in each set period of time
The first transition probability in wheel paths between each two historical position point, above-mentioned set period of time were divided based on a consecutive days
's.
Specifically, due to user drive trip may be related with itself behavioural habits, trip in different time periods may
There are certain regularity, for example, for a working clan user, workaday morning peak section is typically from returning home toward public affairs
Department, and the evening peak period is typically to be gone back home by company, therefore, in order to enable destination prediction model it is more accurate, it is necessary to
With reference to the time that each historical behavior track occurs, the second transfer network of acquisition is pre-processed.
Further, the first transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, above-mentioned
Dimension represents each historical position point before being shifted in above-mentioned goal displacement network, and above-mentioned second dimension represents above-mentioned mesh
Each historical position point after being shifted in mark transfer network, third dimension expression set period of time set, above-mentioned first turn
It moves tensor to be used to indicate in each set period of time, each two historical position point in each history wheelpath of above-mentioned vehicle
Between the first transition probability.
In the preferred embodiment of the present invention, set period of time set is when being divided into several settings by a consecutive days
Between section form.
For example, still illustrated with the example in step 100 and step 101, as shown in fig.2, by the first transport tensor
It is denoted as A ', and A '=(V1, V2, T '), wherein, V1Represent n transient state history bit before being shifted in above-mentioned goal displacement network
It puts a little and n absorbs historical position point, V2Represent n transient state historical position point after being shifted in above-mentioned goal displacement network
Historical position point is absorbed with n, if a consecutive days are divided into h set period of time, i.e. set period of time set T ' is included
H set period of time t, it is assumed that one of entry A ' (i, j, k)=e of the first transport tensor A ', then it represents that historical position point
viAnd vjIn set period of time tk(e.g., 10:00-10:30) transition probability between is e.
Further, due to the history wheelpath closer with current time of origin, more can be current close to user
Driving habit, therefore, in order to make destination prediction model more accurate, according to default time cut-point, by above-mentioned each item
History wheelpath is divided into history wheelpath and recent history wheelpath at a specified future date, i.e. draws above-mentioned goal displacement network
It is divided into long-term goal transfer network and immediate objective transfer network, correspondingly, the first transport tensor can also be divided into long term first
Transport tensor and recent first transport tensor.
Specifically, the building process of the first transport tensor at a specified future date and recent first transport tensor difference is as follows:
Network is shifted according to above-mentioned long-term goal, and at a specified future date the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of above-mentioned first transport tensor at a specified future date represents to occur in above-mentioned long-term goal transfer network
Each historical position point before transfer, the second dimension of above-mentioned first transport tensor at a specified future date represent above-mentioned long-term goal transfer network
It is middle shift after each historical position point, the third dimension of above-mentioned first transport tensor at a specified future date represents above-mentioned set period of time
Set, above-mentioned first transport tensor at a specified future date are used to indicate in each set period of time before above-mentioned time cut-point, on
State the first transition probability of long term between each two historical position point in each history wheelpath of vehicle.
Network is shifted according to above-mentioned immediate objective, and recent the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of above-mentioned recent first transport tensor represents to occur in above-mentioned immediate objective transfer network
Each historical position point before transfer, the second dimension of above-mentioned recent first transport tensor represent above-mentioned immediate objective transfer network
It is middle shift after each historical position point, the third dimension of above-mentioned recent first transport tensor represents above-mentioned set period of time
Set, above-mentioned recent first transport tensor are used to indicate in each set period of time after above-mentioned time cut-point, on
State recent first transition probability between each two historical position point in each history wheelpath of vehicle.
Above-mentioned first transport tensor is formed using above-mentioned first transport tensor at a specified future date and above-mentioned recent first transport tensor.
In the embodiment of the present invention, preferred time cut-point is 20%, i.e. is occurred according to above-mentioned each history wheelpath
Time tandem be ranked up, close to current time history wheelpath preceding, history wheelpath rearward rear,
The goal displacement network representation that each history wheelpath by preceding 20% is formed shifts network for immediate objective, will be by being left
80% each history wheelpath composition goal displacement network representation for long-term goal shift network.
For example, it is assumed that each history wheelpath of above-mentioned vehicle is denoted as R, and by above-mentioned each history wheelpath R again
It is divided into history wheelpath R at a specified future datemWith recent history wheelpath Rc, then, the first transport tensor A ' of acquisition can be divided into far
Phase the first transport tensor A 'mWith recent first transport tensor A 'c, still illustrated above exemplified by an example, if remote being divided into
When the first transport tensor of phase and recent first transport tensor, set period of time set T ' contains h set period of time, then,
After the first transport tensor at a specified future date and recent first transport tensor is divided into, when set period of time set T contains 2h setting
Between section t.
So far, terminal completes the structure to the first transport tensor A ', wherein, it can divide again in the first constructed transport tensor
For the first transport tensor at a specified future date and recent first transport tensor.
Step 104:Terminal obtains central tensor.
Specifically, due to actually calculate when, it is non-from the obtained effective transition probability in above-mentioned first transport tensor
Often few, there are Sparse sex chromosome mosaicisms, therefore, in order to solve Sparse sex chromosome mosaicism, are obtained respectively in above-mentioned goal displacement net
The vehicle fleet size that occurs in network in each set period of time and it is located at each historical position in above-mentioned goal displacement network
Specified point of interest quantity in point preset range forms central tensor.
Further, the vehicle fleet size occurred in above-mentioned goal displacement network in each set period of time, represents thick
Correlation between the different set period captured under the transportation condition of granularity is a statistical value, can directly be obtained from the external world
It takes, the specified point of interest quantity being located in above-mentioned goal displacement network in any one historical position point preset range represents
The specified point of interest quantity that any one above-mentioned historical position point nearby occurs, wherein, specified point of interest can be market, medicine
Shop, hospital, school etc., nearby can also specifically be set as be located at 100 meters of historical position point in the range of, according to actual needs into
Row adjustment, does not limit herein.
Step 105:Central tensor of the terminal based on acquisition, is filled above-mentioned first transport tensor, obtains second turn
Move tensor.
Specifically, after terminal obtains central tensor, each set period of time in above-mentioned goal displacement network is utilized respectively
The specified interest being located in the vehicle fleet size of interior appearance and above-mentioned goal displacement network in each historical position point preset range
The central tensor of point quantity composition, and above-mentioned first transport tensor is filled, obtain the second transport tensor.
Assuming that center tensor representation is obtained as S, the vehicle occurred in above-mentioned goal displacement network in each set period of time
Quantity can be represented with matrix X, be can be used for intensive filling, be located at each historical position point in above-mentioned goal displacement network
Specified point of interest quantity in preset range can be represented with matrix Y, by actual checking computations, by the way that above-mentioned central tensor S is divided
The fac-tor of solution helps to reduce the error of each transition probability included in above-mentioned first transport tensor.
In the embodiment of the present invention, preferably, the following formula, which may be employed, obtains the second transport tensor:
A "=S × V1V1×V2V2×TT
Wherein, A " represents the second transport tensor, V1Represent that n transient state before being shifted in above-mentioned goal displacement network is gone through
History location point and n absorption historical position point, V2Represent n transient state history bit after being shifted in above-mentioned goal displacement network
It puts a little and n absorption historical position point, T represents the set period of time set of above-mentioned first transport tensor, S represents above-mentioned center
Amount.
So far, terminal completes the filling to above-mentioned first transport tensor, obtains the second transport tensor A ".
Step 106:Second transport tensor that terminal-pair obtains optimizes training, obtains each optimization training knot
Fruit filters out the optimization training result for meeting preset rules as goal displacement tensor, and based on above-mentioned goal displacement tensor group
Build destination prediction model.
Specifically, the penalty values between above-mentioned first transport tensor and above-mentioned second transport tensor are calculated, based on above-mentioned damage
Mistake value and each regularization term are fitted above-mentioned second transport tensor, obtain each optimization training result, based on above-mentioned each
A optimization training result is determined for compliance with the optimization training result of preset rules as goal displacement tensor, wherein, above-mentioned each canonical
It is by each historical position point in above-mentioned set period of time set, each history wheelpath of above-mentioned vehicle, above-mentioned to change item
Goal displacement network, the above-mentioned central default geographical feature factor of tensor sum generate.
In the embodiment of the present invention, preferably, following algorithm (gradient descent algorithm), which may be employed, optimizes training, to obtain
Obtain goal displacement tensor:
Wherein, Ω (S, V1,V2, T, G, F) specific calculation it is as follows:
Wherein, A ' and A " is respectively the first transport tensor and the second transport tensor, and L (A', A ") is the loss letter of A ' and A "
Number, can calculate the penalty values between A ' and A ", and S represents above-mentioned central tensor, V1It represents to occur in above-mentioned goal displacement network
N transient state historical position point and n absorption historical position point before transfer, V2It represents to shift in above-mentioned goal displacement network
N transient state historical position point and n absorption historical position point afterwards, T represent the set period of time collection of above-mentioned first transfer network
It closes, G represents lattice number, and F represents the default geographical feature factor (e.g., the dimension of geographical feature), λ4It is first parameter to control
The contribution of different piece, above-mentioned Ω (S, V1,V2, T, G, F) and it is regularization term, prevent overfitting.
In above-mentioned optimization training process, multiple optimization training results can be obtained, be according to gradient in the embodiment of the present invention
Descent algorithm optimizes trained, therefore, according to gradient descent algorithm, chooses minimum in above-mentioned multiple optimization training results
One optimizes training result as goal displacement tensor A, wherein, above-mentioned goal displacement tensor A has recorded each setting time
In section, in each history wheelpath of the vehicle, the goal displacement probability between each two historical position point, above-mentioned every two
A historical position point includes transient state historical position point and absorbs historical position point, and the calculating process of specific transport tensor A is as follows:
Specifically, based on time dimension, the goal displacement tensor of acquisition is split, to obtain each set period of time
Unit transfer matrix, wherein, each element P in each set period of time unit transfer matrixk(i, j) is represented, in tkIf
It fixes time under section, from historical position point viTo historical position point vjTransition probability (viAnd vjCan be transient state historical position point,
Or absorb historical position point).
In the embodiment of the present invention, it is assumed that there are a set period of time unit (set period of time tk) transfer matrix Pk, and will
All absorption historical position points are expressed as Z, and all transient state historical position points represent W, preferably, can be recorded by table 1
Form is by set period of time tkEach transfer matrix PkIt is recombinated:
Table 1
Wherein, I ties up unit matrix for n, and 0 ties up null matrix for a n, and Q is turned between the expression transient state node of a n dimension
The matrix of probability is moved, S is the expression of a n dimension from transient state node to the matrix of transition probability between absorption node.
In the embodiment of the present invention, preferably, the following formula, which may be employed, obtains a set period of time unit (assuming that setting
Period is tk) transfer matrix Pk:
It is specific to represent as follows:
Work as viIt is v0To vn-1One of historical position point, vjIt is v0To vn-1One of historical position point (that is, vi
And vjIt is transient state historical position point) when, viTo vjBetween transition probability be PQkOne of element of (i, j);
Work as viIt is v0To vn-1One of historical position point, vjIt is vnTo v2n-1One of historical position point (i.e.,
viFor transient state historical position point, vjTo absorb historical position point) when, viTo vjBetween transition probability beWherein one
A element;
Work as viIt is vnTo v2n-1One of historical position point, vjIt is v0To vn-1One of historical position point (i.e.,
viTo absorb historical position point, vjFor transient state historical position point) when, viTo vjBetween transition probability be 0 matrix one of them
Element (transient state historical position point can not possibly be then transferred to by absorbing historical position point);
Work as viIt is vnTo v2n-1One of historical position point, vjIt is vnTo v2n-1One of historical position point (i.e.,
viAnd vjIt is to absorb historical position point) when, viTo vjBetween transition probability be unit matrix one of element.
Specifically, in the embodiment of the present invention, preferably, the following formula, which may be employed, obtains PQk(i, j) set period of time tk
Transition probability between interior each two transient state historical position point:
Specifically, in the embodiment of the present invention, preferably, the following formula acquisition may be employedSet period of time tkIt is interior
Each transient state historical position point extremely absorbs transition probability between historical position point:
Wherein, in above-mentioned two calculation formula, | Rm(i, k) | it represents in set period of time tkIt is interior all by viIt is remote
The quantity of phase history wheelpath;|Rc(i, k) | it represents in set period of time tkIt is interior all by viRecent history wheelpath
Quantity;|Rm(i, j, k) | it represents in set period of time tkIt is interior all by viTo vj, and viTo vjBetween be not present a history bit
Put the quantity of history wheelpath at a specified future date a little;|Rc(i, j, k) | it represents in set period of time tkIt is interior all by viTo vj, and vi
To vjBetween there is no historical position point recent history wheelpath quantity;Represent | Rm(i, j, k) | table
In all history wheelpaths at a specified future date shown, terminating point vjHistory wheelpath at a specified future date quantity;Represent | Rc
(i, j, k) | in all recent history wheelpaths of expression, all terminating points are vjRecent history wheelpath quantity.
Above-mentioned calculating process is with set period of time tkExemplified by calculated, for the set period of time set T ',
The calculating of any one set period of time can be according to set period of time tkCalculation carry out, obtain each setting time
In section, the goal displacement probability in each history wheelpath of above-mentioned vehicle between each two historical position point.
So far, by above-mentioned steps, goal displacement tensor is obtained, and based on above-mentioned goal displacement tensor, is completed to purpose
The establishment of ground prediction model, wherein, above-mentioned goal displacement tensor is had recorded respectively in each set period of time, above-mentioned vehicle
In each history wheelpath, each two historical position point (can be transient state historical position point, or absorb historical position
Point) between goal displacement probability, i.e. had recorded respectively in each set period of time in above destination prediction model, institute
In each history wheelpath for stating vehicle, the goal displacement probability between each two historical position point.
In vehicle travel process, terminal is according to established destination prediction model, by the current line for monitoring vehicle
Wheel paths predict the destination in front in time.
Specifically, as shown in fig.3, the detailed process of the destination in front of predictably terminal imagination is as follows:
Step 300:The current driving track of terminal monitoring vehicle, and determine the above-mentioned current driving track corresponding time
Section.
Specifically, terminal in the operational process of vehicle, monitors the current driving track of vehicle, and determine above-mentioned current line
The wheel paths corresponding period, wherein, above-mentioned current driving track contains one group of current location point, and determines that above-mentioned one group is worked as
Each period occurs for front position point.
For example, it is assumed that the current driving track for the vehicle that terminal monitoring arrives is Rp, determine RpComprising one group of current location point
For { vs, vi..., vi+j, vc, if vsAnd vcRespectively current driving track RpInitial position and current stop place point, and
vi、vi+jAnd viBetween vi+jAll current location points be current driving track R respectivelypEach pass by location point, it is determined that on
State one group of current location point { vs, vi..., vi+j, vcOccur each period correspond to { t respectivelys, ti..., ti+j, tc}。
Step 301:Terminal obtains default destination prediction model.
Specifically, terminal is determined according to the current driving track corresponding period got in destination prediction model
In, within the corresponding period of above-mentioned current driving track, the source location set of above-mentioned current driving track, current stop place
Point passes by location point with each, each the mesh between any one above-mentioned historical position point in above-mentioned each history wheelpath
Mark transition probability.
For example, it is assumed that in above-mentioned each history wheelpath with a historical position point vkIf current driving track is still
For the R in step 300p, then, in above destination prediction model, R is determinedpComprising one group of current location point { vs, vi...,
vi+j, vc, within corresponding period, respectively with historical position point vkBetween goal displacement probability.
Step 302:Terminal is based on the above-mentioned current driving track corresponding period, and mould is predicted using the destination of acquisition
Type, the prediction calculated respectively in above-mentioned current driving track and each history wheelpath between each historical position point are general
Rate.
Specifically, terminal determined current driving track and after the corresponding period of above-mentioned current driving track, using purpose
Each the goal displacement probability obtained in ground prediction model is determined in above-mentioned current driving track and each history wheelpath
Prediction probability between each historical position point.
In embodiments of the present invention, preferred embodiment is to be calculated using Bayes's Conditional Probability Computing Method above-mentioned
Prediction probability in current driving track and each history wheelpath between each historical position point.
For example, still with a historical position point v in above-mentioned each history wheelpathkExemplified by illustrate, in pattra leaves
During this conditional probability calculates, current driving track is given, is calculated under current driving track, historical position point vkIt is destination
Prediction probability P (vk|Rp), in the embodiment of the present invention, obtain under current driving track, go through preferably, the following formula may be employed
History location point vkIt is the prediction probability P (v of destinationk|Rp):
P(vk|Rp)
=P (vk|vs,vi,...,vi+j,vc)
=P (vk|vs)·P(vk,vs|vi)·...·P(vk,vs,vi,...,vi+j-1|vi+j)·P(vk,vs,vi,...,
vi+j|vc)
=Bs(s,k)·[Bi(i,k)·PQs(s,i)]·...·[Bi+j(i+j,k)·PQ(i+j-1)(i+j-1,i+j)]·
[Bc(c,k)·PQ(i+j)(i+j,c)]
Wherein, Bs(s, k) is represented in set period of time tsFrom transient state historical position point vsTo absorption historical position point vkSuction
Receive probability, Bi(i, k) is represented in set period of time tiFrom transient state historical position point viTo absorption historical position point vkAbsorption it is general
Rate, PQs(s, i) is represented in set period of time tsFrom transient state historical position point vsTo transient state historical position point viGoal displacement it is general
Rate, Bi+j(i+j, k) is represented in set period of time ti+jFrom transient state historical position point vi+jTo absorption historical position point vkAbsorption it is general
Rate, PQ(i+j-1)(i+j-1, i+j) is represented in set period of time ti+j-1From transient state historical position point vi+j-1To transient state historical position point
vi+jGoal displacement probability, Bc(c, k) is represented in set period of time tcFrom transient state historical position point vcTo absorption historical position point
vkAbsorbing probability, PQ(i+j)(i+j, c) is represented in set period of time ti+jFrom transient state historical position point vi+jTo transient state historical position
Point vcGoal displacement probability.
Further, above-mentioned each absorbing probability is based on the goal displacement probability and transient state between transient state historical position point
What historical position point and the goal displacement probability absorbed between historical position point obtained, in the embodiment of the present invention, preferably, can adopt
Set period of time t is obtained with the following formulakEach absorbing probability:
Bk=HkSk=(I-Qk)-1Sk
Wherein, QkFor in goal displacement tensor A, in set period of time tkMesh between interior any two transient state historical position point
Mark transition probability, SkFor in goal displacement tensor A, in set period of time tkAny one interior transient state historical position point with it is any one
A goal displacement probability absorbed between historical position point.
BkFor set period of time tkEach absorbing probability, i.e. absorbing probability matrix, B be each set period of time it is each
A absorbing probability, i.e. tensor is absorbed, for set period of time tkAn absorbing probability Bk(i, j) is then represented, in setting time
Section tkFrom transient state historical position point viTo absorption historical position point vjAbsorbing probability.
In above-mentioned calculating process, each goal displacement probability being related to can be beforehand through step 302 in above-mentioned mesh
Ground prediction model in obtain, pass through above-mentioned calculating, it is known that under current driving track, historical position point vkIt is the pre- of destination
Survey probability P (vk|Rp) occurrence.
And so on, under current driving track, other history bits that each history wheelpath of above-mentioned vehicle includes
Put is a little that the prediction probability of destination can also be used above-mentioned calculation and obtain.
So far, terminal is obtained in above-mentioned current driving track and each history wheelpath between each historical position point
Prediction probability.
Step 303:The historical position point that corresponding prediction probability is met preset condition by terminal is determined as predicting destination.
Specifically, each prediction probability that terminal-pair obtains is screened, pair that prediction probability meets preset condition is chosen
The historical position point answered is determined as predicting destination, in the embodiment of the present invention, preferably, each prediction probability of acquisition is carried out
Sequence, using the corresponding historical position point of prediction probability maximum in above-mentioned each prediction probability as prediction destination.
So far, terminal is completed according to established destination prediction model, by monitor vehicle current driving track and
When prediction front destination.
As shown in fig.4, in the embodiment of the present invention, prediction destination device includes at least monitoring unit 41, acquiring unit
42nd, computing unit 43 and determination unit 44, wherein,
Monitoring unit 41 for monitoring the current driving track of vehicle, determines the current driving track corresponding time
Section;
Acquiring unit 42 for obtaining default destination prediction model, records respectively in the destination prediction model
In each set period of time, in each history wheelpath of the vehicle, the target between each two historical position point
Transition probability, wherein, the set period of time was divided based on a consecutive days;
For being based on the current driving track corresponding period, mould is predicted using the destination for computing unit 43
Type, the prediction calculated respectively in the current driving track and each history wheelpath between each historical position point are general
Rate;
Determination unit 44, the historical position point for corresponding prediction probability to be met to preset condition are determined as predicting purpose
Ground.
Optionally, described device further includes:Model foundation unit 45;
The model foundation unit 45 is used for, and before the current driving track of the vehicle is monitored, generates the purpose
Ground prediction model, specifically includes:
Obtain each history wheelpath of the vehicle;
Each history wheelpath of the vehicle of acquisition is projected to road network, obtains goal displacement network,
Wherein, the goal displacement network is used to indicate opposite between each historical position point in each history wheelpath
Movement;
First transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, first dimension represents
Each historical position point before being shifted in the goal displacement network, second dimension represent the goal displacement network
It is middle shift after each historical position point, third dimension represents set period of time set, and first transport tensor is used for
Indicate in each set period of time, in each history wheelpath of the vehicle between each two historical position point first
Transition probability;
First transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, wherein, the mesh
Mark transport tensor is had recorded respectively in each set period of time, and in each history wheelpath of the vehicle, each two is gone through
Goal displacement probability between history location point.
Optionally, each history wheelpath of the vehicle of acquisition is projected to road network, obtains target and turn
When moving network, the model foundation unit 45 is used for:
Each history wheelpath of the vehicle of acquisition is projected to road network, road network in the projected
In establish the first transfer network;
Each historical position point in each history wheelpath included in network is shifted by described first to identify
For transient state historical position point, and the absorption historical position point of a mirror image is created for each transient state historical position point, wherein, institute
The nonterminal point in transient state historical position point expression history wheelpath is stated, the absorption historical position point represents history driving rail
Terminating point in mark;
Based between each transient state historical position point in each history wheelpath and each absorption historical position point
Relative motion, establish corresponding goal displacement network.
Optionally, when establishing the first transport tensor based on the first dimension, the second dimension and third dimension, the model foundation
Unit 45 is used for:
By the goal displacement network according to default time cut-point, long-term goal transfer network and recent mesh are divided into
Mark transfer network, wherein, the long-term goal transfer network includes in the goal displacement network in the time cut-point
The historical position point generated before, the immediate objective transfer network included in the goal displacement network in the time point
The historical position point generated after cutpoint;
Network is shifted according to the long-term goal, and at a specified future date the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of first transport tensor at a specified future date represents to occur in the long-term goal transfer network
Each historical position point before transfer, the second dimension of first transport tensor at a specified future date represent the long-term goal transfer network
It is middle shift after each historical position point, the third dimension of first transport tensor at a specified future date represents the set period of time
Set, first transport tensor at a specified future date are used to indicate in each set period of time before the time cut-point, institute
State the first transition probability of long term between each two historical position point in each history wheelpath of vehicle;
Network is shifted according to the immediate objective, and recent the is established based on the first dimension, the second dimension and third dimension
One transport tensor, wherein, the first dimension of recent first transport tensor represents to occur in the immediate objective transfer network
Each historical position point before transfer, the second dimension of recent first transport tensor represent the immediate objective transfer network
It is middle shift after each historical position point, the third dimension of recent first transport tensor represents the set period of time
Set, recent first transport tensor are used to indicate in each set period of time after the time cut-point, institute
State recent first transition probability between each two historical position point in each history wheelpath of vehicle;
Using first transport tensor at a specified future date and recent first transport tensor, first transport tensor is formed.
Optionally, first transport tensor is handled, when acquisition meets the goal displacement tensor of preset rules, institute
Model foundation unit 45 is stated to be used for:
The vehicle fleet size that occurs in the goal displacement network in each set period of time is obtained respectively and described
The specified point of interest quantity being located in goal displacement network in each historical position point preset range forms central tensor;
The central tensor based on acquisition, is filled first transport tensor, obtains the second transport tensor;
Training is optimized to second transport tensor of acquisition, obtains each optimization training result;
Based on each optimization training result, the optimization training result of preset rules is determined for compliance with as goal displacement
Amount.
Optionally, training is optimized to second transport tensor of acquisition, when obtaining each optimization training result, institute
Model foundation unit 45 is stated to be used for:
Calculate the penalty values between first transport tensor and second transport tensor;
Based on the penalty values and each regularization term, second transport tensor is fitted, obtains each optimization
Training result, wherein, each regularization term is by the set period of time set, each history of vehicle driving rail
Each historical position point, the goal displacement network, the central default geographical feature factor of tensor sum generate in mark.
Optionally, based on the current driving track corresponding period, using the destination prediction model, institute is calculated
When stating in current driving track and each history wheelpath the prediction probability between any one historical position point, the calculating
Unit 43 is used for:
Determine that source location set, current stop place point and the source location set of the current driving track are stopped with current
Each between location point passes by location point;
It determines in the destination prediction model, within the corresponding period of the current driving track, the current line
The source location set of wheel paths, current stop place point and it is each pass by location point, each with each history wheelpath
Described in goal displacement probability between any one historical position point;
Each goal displacement probability based on acquisition determines the current driving track and each history driving rail
Prediction probability described in mark between any one historical position point.
In the embodiment of the present invention, each history wheelpath of vehicle and above-mentioned each history driving rail are in advance based on
The otherness of the time factor of mark, terminating point and nonterminal point, establishes a destination prediction model, wherein, above destination
It is had recorded respectively in each set period of time in prediction model, in each history wheelpath of the vehicle, each two is gone through
Goal displacement probability between history location point when monitoring the current driving track of vehicle, determines above-mentioned current driving track pair
The period answered, and based on above-mentioned current driving track and above-mentioned current track corresponding period, using pre-set mesh
Ground prediction model, calculate respectively in above-mentioned current driving track and each history wheelpath between each historical position point
Prediction probability, by corresponding prediction probability meet preset condition historical position point be determined as predict destination, in this way, can examine
Consider the otherness of nonterminal point and terminating point in historical behavior track, at the same can also fully take into account historical behavior track when
Between factor so that the destination prediction model of foundation is more accurate, so as to improve destination prediction accuracy, and then promoted
The driving experience of user.
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, system or computer program
Product.Therefore, the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware can be used in the present invention
Apply the form of example.Moreover, the computer for wherein including computer usable program code in one or more can be used in the present invention
The computer program production that usable storage medium is implemented on (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention be with reference to according to the method for the embodiment of the present invention, the flow of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that it can be realized by computer program instructions every first-class in flowchart and/or the block diagram
The combination of flow and/or box in journey and/or box and flowchart and/or the block diagram.These computer programs can be provided
The processor of all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that the instruction performed by computer or the processor of other programmable data processing devices is generated for real
The device for the function of being specified in present one flow of flow chart or one box of multiple flows and/or block diagram or multiple boxes.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction generation being stored in the computer-readable memory includes referring to
Make the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one box of block diagram or
The function of being specified in multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted
Series of operation steps is performed on calculation machine or other programmable devices to generate computer implemented processing, so as in computer or
The instruction offer performed on other programmable devices is used to implement in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although preferred embodiments of the present invention have been described, but those skilled in the art once know basic creation
Property concept, then can make these embodiments other change and modification.So appended claims be intended to be construed to include it is excellent
It selects embodiment and falls into all change and modification of the scope of the invention.
Obviously, those skilled in the art can carry out the embodiment of the present invention various modification and variations without departing from this hair
The spirit and scope of bright embodiment.In this way, if these modifications and variations of the embodiment of the present invention belong to the claims in the present invention
And its within the scope of equivalent technologies, then the present invention is also intended to comprising including these modification and variations.
Claims (14)
1. a kind of destination Forecasting Methodology, which is characterized in that including:
The current driving track of vehicle is monitored, determines the current driving track corresponding period;
Default destination prediction model is obtained, each set period of time is had recorded respectively in the destination prediction model
It is interior, in each history wheelpath of the vehicle, the goal displacement probability between each two historical position point, wherein, it is described
Set period of time was divided based on a consecutive days;
Based on the current driving track corresponding period, using the destination prediction model, calculate respectively described current
Prediction probability in wheelpath and each history wheelpath between each historical position point;
The historical position point that corresponding prediction probability is met to preset condition is determined as predicting destination.
2. the method as described in claim 1, which is characterized in that before the current driving track for monitoring the vehicle, further
Including:The destination prediction model is generated, is specifically included:
Obtain each history wheelpath of the vehicle;
Each history wheelpath of the vehicle of acquisition is projected to road network, obtains goal displacement network, wherein,
The goal displacement network is used to indicate the relative motion between each historical position point in each history wheelpath;
First transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, described in the first dimension expression
Each historical position point before being shifted in goal displacement network, second dimension represent to send out in the goal displacement network
Each historical position point after raw transfer, third dimension represent set period of time set, and first transport tensor is used to indicate
In each set period of time, the first transfer in each history wheelpath of the vehicle between each two historical position point
Probability;
First transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, wherein, the target turns
Tensor is moved to have recorded respectively in each set period of time, in each history wheelpath of the vehicle, each two history bit
Goal displacement probability between putting a little.
3. method as claimed in claim 2, which is characterized in that project each history wheelpath of the vehicle of acquisition
To road network, goal displacement network is obtained, including:
Each history wheelpath of the vehicle of acquisition is projected to road network, is built in road network in the projected
Vertical first transfer network;
Each historical position point in each history wheelpath included in network, which is shifted, by described first is identified as wink
State historical position point, and be the absorption historical position point of each transient state historical position point one mirror image of establishment, wherein, the wink
State historical position point represents the nonterminal point in history wheelpath, and the absorption historical position point is represented in history wheelpath
Terminating point;
Based on each transient state historical position point in each history wheelpath and each phase absorbed between historical position point
To movement, corresponding goal displacement network is established.
4. method as claimed in claim 2, which is characterized in that establish based on the first dimension, the second dimension and third dimension
One transport tensor, including:
By the goal displacement network according to default time cut-point, it is divided into long-term goal transfer network and immediate objective turns
Network is moved, wherein, the long-term goal transfer network includes in the goal displacement network before the time cut-point
The historical position point of generation, the immediate objective transfer network include in the goal displacement network in the time cut-point
The historical position point generated afterwards;
Network is shifted according to the long-term goal, and first turn of long term is established based on the first dimension, the second dimension and third dimension
Tensor is moved, wherein, the first dimension of first transport tensor at a specified future date represents to shift in the long-term goal transfer network
Preceding each historical position point, the second dimension of first transport tensor at a specified future date represent to send out in the long-term goal transfer network
Each historical position point after raw transfer, the third dimension of first transport tensor at a specified future date represent the set period of time collection
It closes, first transport tensor at a specified future date is used to indicate in each set period of time before the time cut-point, described
The first transition probability of long term in each history wheelpath of vehicle between each two historical position point;
Network is shifted according to the immediate objective, and recent first turn is established based on the first dimension, the second dimension and third dimension
Tensor is moved, wherein, the first dimension of recent first transport tensor represents to shift in the immediate objective transfer network
Preceding each historical position point, the second dimension of recent first transport tensor represent to send out in the immediate objective transfer network
Each historical position point after raw transfer, the third dimension of recent first transport tensor represent the set period of time collection
It closes, recent first transport tensor is used to indicate in each set period of time after the time cut-point, described
Recent first transition probability in each history wheelpath of vehicle between each two historical position point;
Using first transport tensor at a specified future date and recent first transport tensor, first transport tensor is formed.
5. method as claimed in claim 2, which is characterized in that handle first transport tensor, acquisition meets pre-
If the goal displacement tensor of rule, including:
The vehicle fleet size that occurs in the goal displacement network in each set period of time is obtained respectively and in the target
The specified point of interest quantity being located in network in each historical position point preset range is shifted, forms central tensor;
The central tensor based on acquisition, is filled first transport tensor, obtains the second transport tensor;
Training is optimized to second transport tensor of acquisition, obtains each optimization training result;
Based on each optimization training result, the optimization training result for being determined for compliance with preset rules is goal displacement tensor.
6. method as claimed in claim 5, which is characterized in that training is optimized to second transport tensor of acquisition,
Each optimization training result is obtained, including:
Calculate the penalty values between first transport tensor and second transport tensor;
Based on the penalty values and each regularization term, second transport tensor is fitted, obtains each optimization training
As a result, wherein, each regularization term is by the set period of time set, each history wheelpath of the vehicle
Each historical position point, the goal displacement network, the central default geographical feature factor of tensor sum generate.
7. such as claim 1-6 any one of them methods, which is characterized in that based on the current driving track corresponding time
Section, using the destination prediction model, calculates the current driving track and is gone through with any one in each history wheelpath
Prediction probability between history location point, including:
Determine source location set, current stop place point and the source location set of the current driving track and current stop place
Each location point is passed by between point;
It determines in the destination prediction model, within the corresponding period of the current driving track, the current line track
The source location set of mark, current stop place point and it is each pass by location point, each with institute in each history wheelpath
State the goal displacement probability between any one historical position point;
Each goal displacement probability based on acquisition is determined in the current driving track and each history wheelpath
Prediction probability between any one described historical position point.
8. a kind of destination prediction meanss, which is characterized in that including:
Monitoring unit for monitoring the current driving track of vehicle, determines the current driving track corresponding period;
Acquiring unit for obtaining default destination prediction model, has recorded each respectively in the destination prediction model
In a set period of time, in each history wheelpath of the vehicle, the goal displacement between each two historical position point is general
Rate, wherein, the set period of time was divided based on a consecutive days;
Computing unit, for being based on the current driving track corresponding period, using the destination prediction model, difference
Calculate the prediction probability between each historical position point in the current driving track and each history wheelpath;
Determination unit, the historical position point for corresponding prediction probability to be met to preset condition are determined as predicting destination.
9. device as claimed in claim 8, which is characterized in that described device further includes:Model foundation unit;
The model foundation unit is used for, and before the current driving track of the vehicle is monitored, generates the destination prediction
Model specifically includes:
Obtain each history wheelpath of the vehicle;
Each history wheelpath of the vehicle of acquisition is projected to road network, obtains goal displacement network, wherein,
The goal displacement network is used to indicate the relative motion between each historical position point in each history wheelpath;
First transport tensor is established based on the first dimension, the second dimension and third dimension, wherein, described in the first dimension expression
Each historical position point before being shifted in goal displacement network, second dimension represent to send out in the goal displacement network
Each historical position point after raw transfer, third dimension represent set period of time set, and first transport tensor is used to indicate
In each set period of time, the first transfer in each history wheelpath of the vehicle between each two historical position point
Probability;
First transport tensor is handled, obtains the goal displacement tensor for meeting preset rules, wherein, the target turns
Tensor is moved to have recorded respectively in each set period of time, in each history wheelpath of the vehicle, each two history bit
Goal displacement probability between putting a little.
10. device as claimed in claim 9, which is characterized in that throw each history wheelpath of the vehicle of acquisition
After shadow to road network, when obtaining goal displacement network, the model foundation unit is used for:
Each history wheelpath of the vehicle of acquisition is projected to road network, is built in road network in the projected
Vertical first transfer network;
Each historical position point in each history wheelpath included in network, which is shifted, by described first is identified as wink
State historical position point, and be the absorption historical position point of each transient state historical position point one mirror image of establishment, wherein, the wink
State historical position point represents the nonterminal point in history wheelpath, and the absorption historical position point is represented in history wheelpath
Terminating point;
Based on each transient state historical position point in each history wheelpath and each phase absorbed between historical position point
To movement, corresponding goal displacement network is established.
11. device as claimed in claim 9, which is characterized in that establish based on the first dimension, the second dimension and third dimension
During one transport tensor, the model foundation unit is used for:
By the goal displacement network according to default time cut-point, it is divided into long-term goal transfer network and immediate objective turns
Network is moved, wherein, the long-term goal transfer network includes in the goal displacement network before the time cut-point
The historical position point of generation, the immediate objective transfer network include in the goal displacement network in the time cut-point
The historical position point generated afterwards;
Network is shifted according to the long-term goal, and first turn of long term is established based on the first dimension, the second dimension and third dimension
Tensor is moved, wherein, the first dimension of first transport tensor at a specified future date represents to shift in the long-term goal transfer network
Preceding each historical position point, the second dimension of first transport tensor at a specified future date represent to send out in the long-term goal transfer network
Each historical position point after raw transfer, the third dimension of first transport tensor at a specified future date represent the set period of time collection
It closes, first transport tensor at a specified future date is used to indicate in each set period of time before the time cut-point, described
The first transition probability of long term in each history wheelpath of vehicle between each two historical position point;
Network is shifted according to the immediate objective, and recent first turn is established based on the first dimension, the second dimension and third dimension
Tensor is moved, wherein, the first dimension of recent first transport tensor represents to shift in the immediate objective transfer network
Preceding each historical position point, the second dimension of recent first transport tensor represent to send out in the immediate objective transfer network
Each historical position point after raw transfer, the third dimension of recent first transport tensor represent the set period of time collection
It closes, recent first transport tensor is used to indicate in each set period of time after the time cut-point, described
Recent first transition probability in each history wheelpath of vehicle between each two historical position point;
Using first transport tensor at a specified future date and recent first transport tensor, first transport tensor is formed.
12. device as claimed in claim 9, which is characterized in that handle first transport tensor, acquisition meets pre-
If during the goal displacement tensor of rule, the model foundation unit is used for:
The vehicle fleet size that occurs in the goal displacement network in each set period of time is obtained respectively and in the target
The specified point of interest quantity being located in network in each historical position point preset range is shifted, forms central tensor;
The central tensor based on acquisition, is filled first transport tensor, obtains the second transport tensor;
Training is optimized to second transport tensor of acquisition, obtains each optimization training result;
Based on each optimization training result, the optimization training result for being determined for compliance with preset rules is goal displacement tensor.
13. device as claimed in claim 12, which is characterized in that instruction is optimized to second transport tensor of acquisition
Practice, when obtaining each optimization training result, the model foundation unit is used for:
Calculate the penalty values between first transport tensor and second transport tensor;
Based on the penalty values and each regularization term, second transport tensor is fitted, obtains each optimization training
As a result, wherein, each regularization term is by the set period of time set, each history wheelpath of the vehicle
Each historical position point, the goal displacement network, the central default geographical feature factor of tensor sum generate.
14. such as claim 8-13 any one of them devices, which is characterized in that based on the current driving track it is corresponding when
Between section, using the destination prediction model, calculate the current driving track and any one in each history wheelpath
During prediction probability between historical position point, the computing unit is used for:
Determine source location set, current stop place point and the source location set of the current driving track and current stop place
Each location point is passed by between point;
It determines in the destination prediction model, within the corresponding period of the current driving track, the current line track
The source location set of mark, current stop place point and it is each pass by location point, each with institute in each history wheelpath
State the goal displacement probability between any one historical position point;
Each goal displacement probability based on acquisition is determined in the current driving track and each history wheelpath
Prediction probability between any one described historical position point.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611050832.XA CN108108831B (en) | 2016-11-24 | 2016-11-24 | Destination prediction method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611050832.XA CN108108831B (en) | 2016-11-24 | 2016-11-24 | Destination prediction method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108108831A true CN108108831A (en) | 2018-06-01 |
CN108108831B CN108108831B (en) | 2020-12-04 |
Family
ID=62204910
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611050832.XA Active CN108108831B (en) | 2016-11-24 | 2016-11-24 | Destination prediction method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108108831B (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109389246A (en) * | 2018-09-13 | 2019-02-26 | 中国科学院电子学研究所苏州研究院 | A kind of vehicles destination regional scope prediction technique neural network based |
CN109466479A (en) * | 2018-10-26 | 2019-03-15 | 广州小鹏汽车科技有限公司 | Control method for vehicle, device, terminal device and medium |
CN110161418A (en) * | 2019-06-03 | 2019-08-23 | 北京中科奥创科技有限公司 | Health degree prediction technique, system and the computer readable storage medium of battery |
WO2019232776A1 (en) * | 2018-06-08 | 2019-12-12 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for generating personalized destination recommendations |
CN110909106A (en) * | 2019-11-26 | 2020-03-24 | 杭州三汇数字信息技术有限公司 | Trajectory prediction method and system |
CN111131999A (en) * | 2018-10-15 | 2020-05-08 | 中国移动通信集团有限公司 | Method and device for predicting destination based on position track |
CN111179589A (en) * | 2019-12-06 | 2020-05-19 | 北京中交兴路信息科技有限公司 | Method, device, equipment and storage medium for predicting vehicle OD |
CN112749740A (en) * | 2020-12-30 | 2021-05-04 | 北京优挂信息科技有限公司 | Method, device, electronic equipment and medium for determining vehicle destination |
CN113379304A (en) * | 2021-06-28 | 2021-09-10 | 北京百度网讯科技有限公司 | Method and device for determining electricity consumption |
CN113830098A (en) * | 2021-11-12 | 2021-12-24 | 比亚迪股份有限公司 | Vehicle driving reminding method and device, storage medium and vehicle |
CN114169589A (en) * | 2021-11-22 | 2022-03-11 | 北京中交兴路信息科技有限公司 | Method, device, storage medium and terminal for predicting vehicle destination |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104408203A (en) * | 2014-12-18 | 2015-03-11 | 西安电子科技大学宁波信息技术研究院 | Method for predicting path destination of moving object |
CN104850604A (en) * | 2015-05-04 | 2015-08-19 | 华中科技大学 | Tensor-based user track mining method |
CN105760958A (en) * | 2016-02-24 | 2016-07-13 | 电子科技大学 | Vehicle track prediction method based on Internet of vehicles |
US20160210860A1 (en) * | 2013-06-26 | 2016-07-21 | Bayerische Motoren Werke Aktiengesellschaft | Method for processing measurement data of a vehicle in order to determine the start of a search for a parking space |
-
2016
- 2016-11-24 CN CN201611050832.XA patent/CN108108831B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160210860A1 (en) * | 2013-06-26 | 2016-07-21 | Bayerische Motoren Werke Aktiengesellschaft | Method for processing measurement data of a vehicle in order to determine the start of a search for a parking space |
CN104408203A (en) * | 2014-12-18 | 2015-03-11 | 西安电子科技大学宁波信息技术研究院 | Method for predicting path destination of moving object |
CN104850604A (en) * | 2015-05-04 | 2015-08-19 | 华中科技大学 | Tensor-based user track mining method |
CN105760958A (en) * | 2016-02-24 | 2016-07-13 | 电子科技大学 | Vehicle track prediction method based on Internet of vehicles |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11011057B2 (en) | 2018-06-08 | 2021-05-18 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for generating personalized destination recommendations |
WO2019232776A1 (en) * | 2018-06-08 | 2019-12-12 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for generating personalized destination recommendations |
CN112236787B (en) * | 2018-06-08 | 2024-06-11 | 北京嘀嘀无限科技发展有限公司 | System and method for generating personalized destination recommendations |
CN109389246B (en) * | 2018-09-13 | 2021-03-16 | 中国科学院电子学研究所苏州研究院 | Neural network-based vehicle destination area range prediction method |
CN109389246A (en) * | 2018-09-13 | 2019-02-26 | 中国科学院电子学研究所苏州研究院 | A kind of vehicles destination regional scope prediction technique neural network based |
CN111131999A (en) * | 2018-10-15 | 2020-05-08 | 中国移动通信集团有限公司 | Method and device for predicting destination based on position track |
CN111131999B (en) * | 2018-10-15 | 2021-05-25 | 中国移动通信集团有限公司 | Method and device for predicting destination based on position track |
CN109466479A (en) * | 2018-10-26 | 2019-03-15 | 广州小鹏汽车科技有限公司 | Control method for vehicle, device, terminal device and medium |
CN110161418A (en) * | 2019-06-03 | 2019-08-23 | 北京中科奥创科技有限公司 | Health degree prediction technique, system and the computer readable storage medium of battery |
CN110909106A (en) * | 2019-11-26 | 2020-03-24 | 杭州三汇数字信息技术有限公司 | Trajectory prediction method and system |
CN110909106B (en) * | 2019-11-26 | 2022-08-12 | 杭州三汇数字信息技术有限公司 | Trajectory prediction method and system |
CN111179589A (en) * | 2019-12-06 | 2020-05-19 | 北京中交兴路信息科技有限公司 | Method, device, equipment and storage medium for predicting vehicle OD |
CN111179589B (en) * | 2019-12-06 | 2021-06-29 | 北京中交兴路信息科技有限公司 | Method, device, equipment and storage medium for predicting vehicle OD |
CN112749740A (en) * | 2020-12-30 | 2021-05-04 | 北京优挂信息科技有限公司 | Method, device, electronic equipment and medium for determining vehicle destination |
CN113379304A (en) * | 2021-06-28 | 2021-09-10 | 北京百度网讯科技有限公司 | Method and device for determining electricity consumption |
CN113830098A (en) * | 2021-11-12 | 2021-12-24 | 比亚迪股份有限公司 | Vehicle driving reminding method and device, storage medium and vehicle |
CN114169589A (en) * | 2021-11-22 | 2022-03-11 | 北京中交兴路信息科技有限公司 | Method, device, storage medium and terminal for predicting vehicle destination |
Also Published As
Publication number | Publication date |
---|---|
CN108108831B (en) | 2020-12-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108108831A (en) | A kind of destination Forecasting Methodology and device | |
US10558898B2 (en) | System and method for generating a map from activity data | |
Yao et al. | Parallel hyper-heuristic algorithm for multi-objective route planning in a smart city | |
AU2013205716B2 (en) | Location-based cognitive and predictive communication system | |
US9671237B1 (en) | Optimized generation of navigation instructions based on computed parking probability values | |
EP3567539A1 (en) | Method and system for orchestrating multi-party services using semi-cooperative nash equilibrium based on artificial intelligence, neural network models, reinforcement learning and finite-state automata | |
CN105091889B (en) | A kind of determination method and apparatus of hotspot path | |
CN105551244B (en) | A kind of dynamic path planning method | |
CN103945486B (en) | Multiple-objection optimization data transferring method based on mobile node track | |
CN103049805A (en) | Vehicle route optimization method with time window constraint based on improved particle swarm optimization (PSO) | |
CN103063222B (en) | With the navigation system and its operating method for turning to restriction scheme | |
US20230259873A1 (en) | Delivery plan generation device and delivery plan generation method | |
CN107949873A (en) | Hourage data point reuse device, hourage data adjustment method and program | |
CN103793763A (en) | Optimal bus taking route excavating system based on big data and cloud computing | |
CN112163701B (en) | Station hub transfer management method and device | |
CN109828579A (en) | A kind of method for planning path for mobile robot that target delta is mobile | |
CN105809290A (en) | Method and device for realizing logistic scheduling | |
CN103039036A (en) | Method and system for calculating number of users | |
Kong et al. | Mobility trajectory generation: a survey | |
CN114723480A (en) | Passenger flow prediction method and cargo scheduling system for rural tourism | |
CN118195341A (en) | Regional carbon emission prediction method and device based on road network and travel demand data | |
CN107576332A (en) | A kind of method and apparatus of transfering navigation | |
Li | Research on scheduling algorithm of agricultural machinery cooperative operation based on particle swarm neural network | |
JP2015076069A (en) | Planar graph generation device, program and method | |
KR101466668B1 (en) | Location Classifying method and electronic apparatus for the same 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 | ||
CP01 | Change in the name or title of a patent holder | ||
CP01 | Change in the name or title of a patent holder |
Address after: 100053 53a, xibianmennei street, Xuanwu District, Beijing Patentee after: CHINA MOBILE COMMUNICATION LTD., Research Institute Patentee after: CHINA MOBILE COMMUNICATIONS GROUP Co.,Ltd. Address before: 100053 53a, xibianmennei street, Xuanwu District, Beijing Patentee before: CHINA MOBILE COMMUNICATION LTD., Research Institute Patentee before: CHINA MOBILE COMMUNICATIONS Corp. |