CN112700639B - Intelligent traffic path planning method based on federal learning and digital twins - Google Patents
Intelligent traffic path planning method based on federal learning and digital twins Download PDFInfo
- Publication number
- CN112700639B CN112700639B CN202011418422.2A CN202011418422A CN112700639B CN 112700639 B CN112700639 B CN 112700639B CN 202011418422 A CN202011418422 A CN 202011418422A CN 112700639 B CN112700639 B CN 112700639B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- model
- vehicles
- indicating
- digital twin
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 230000002776 aggregation Effects 0.000 claims abstract description 11
- 238000004220 aggregation Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 11
- 230000004931 aggregating effect Effects 0.000 claims abstract description 4
- 230000000977 initiatory effect Effects 0.000 claims abstract description 4
- 238000004891 communication Methods 0.000 claims description 25
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000006116 polymerization reaction Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 4
- 230000006855 networking Effects 0.000 abstract description 6
- 230000009286 beneficial effect Effects 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0968—Systems involving transmission of navigation instructions to the vehicle
- G08G1/096805—Systems involving transmission of navigation instructions to the vehicle where the transmitted instructions are used to compute a route
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W12/00—Security arrangements; Authentication; Protecting privacy or anonymity
- H04W12/06—Authentication
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Security & Cryptography (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Remote Sensing (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses an intelligent traffic path planning method based on federal learning and digital twins, which comprises the following steps: s1: vehicle registration is carried out in a vehicle system, and vehicle identity information is verified; s2: according to local historical data of the vehicle, participating in federal learning and training a local model; s3: aggregating local models of all vehicles to obtain an aggregate model; s4: judging whether the aggregation model converges to the preset precision or exceeds the time limit, if so, entering the step S5, otherwise, returning to the step S2; s5: establishing a global digital twin model of the Internet of vehicles; s6: periodically updating the global digital twin model of the Internet of vehicles; s7: and initiating a routing request to the roadside unit, and updating the optimal path and the local prediction model in real time. The planning method is applied to the field of vehicle networking and aims to solve the problems that in the current road traffic system, the flow prediction and path planning are low in accuracy, high in time delay and prone to privacy disclosure.
Description
Technical Field
The invention belongs to the technical field of vehicle networking, and particularly relates to an intelligent traffic path planning method based on federal learning and digital twins.
Background
In real life, digital twins are deeply combined with artificial intelligence technology, real-time interaction and fusion of information space and physical space are promoted, so that more real digital simulation is carried out in an information platform, and wider application is realized. The digital twin system is combined with machine learning framework learning, and the digital twin system can perform self-learning according to multiple feedback source data, so that the real condition of a physical entity is presented in the digital world almost in real time, and the impending events can be presumed and previewed.
However, the digital twinning technique also faces many challenges. First, the digital twin mirroring of the network requires the periodic collection of large data tables from the devices. Secondly, the virtual mirror of the network needs to communicate with the physical network frequently to ensure the real-time performance of the information. In addition, the wireless communication link may be subject to much interference, which may result in problems such as long data transmission time.
However, the following problems are prevalent in the existing studies: (1) due to the fact that the topology of the internet of vehicles is complex and various in changes, and the communication state is unstable and the bandwidth is short, the current scheme cannot collect the running state and road condition information of the vehicles in real time, so that an existing decision mechanism is difficult to make effective decisions in real time according to the dynamic changes of network parameters; (2) due to the storage limit and the communication range limit of the roadside unit, the data scale of decision-making is limited, and the accuracy of the prediction model needs to be improved. The existing scheme lacks research on a larger-scale collaborative decision mechanism, so that the decision accuracy rate is difficult to meet the requirement; (3) there is a high risk of privacy disclosure to the data generated by the vehicle and the user. Since data generated in the internet of vehicles may contain user sensitive information, the existing solutions have a high risk of leakage during data transmission and processing.
Therefore, how to establish a set of safe, reliable, intelligent and efficient intelligent traffic decision-making mechanism aiming at the characteristics of dynamic changes of the internet of vehicles is a research focus of the invention, and the data privacy of the user is protected while the decision-making efficiency and the accuracy are improved.
Disclosure of Invention
The invention aims to solve the problem of establishing a safe, reliable, intelligent and efficient intelligent traffic decision mechanism, and provides an intelligent traffic path planning method based on federal learning and digital twins.
The technical scheme of the invention is as follows: an intelligent traffic path planning method based on federal learning and digital twins comprises the following steps:
s1: vehicle registration is carried out in a vehicle system, and vehicle identity information is verified;
s2: passing vehicle c according to verificationrLocal history data D ofiParticipating in federal learning and training local model mi;
S3: using roadside units RkEdge server of (2) aggregating all vehicles crLocal model m ofiTo obtain a polymerization model Mk;
S4: repeating the steps S2-S3 to judge the aggregation model MkWhether to converge to a predetermined accuracy thetathOr exceeding the time limit TthIf yes, go to step S5, otherwise return to step S2;
s5: updating vehicles crOf real-time local State S'iTo roadside unit RkAnd using a plurality of roadside units RkEstablishing a global digital twin model G (t) of the Internet of vehicles;
s6: using roadside units RkPeriodically updating the global digital twin model G (t) of the Internet of vehicles;
s7: according to the updated global digital twin model G (t) of the Internet of vehicles, utilizing the vehicle crTo roadside unit RkInitiating a way finding request reqr,iAnd updating the optimal path and the local prediction model in real time to complete the optimal path planning of the intelligent traffic.
The invention has the beneficial effects that:
(1) the planning method is applied to the field of vehicle networking and aims to solve the problems that in the current road traffic system, the flow prediction and path planning are low in accuracy, high in time delay and prone to privacy disclosure. In the invention, in the process of traffic flow optimization control, by introducing a vehicle digital twin network model, the process of analyzing and calculating traffic flow data is effectively transferred to the roadside unit edge layer, thereby avoiding the risks brought by unreliable communication states and limited calculation resources of the vehicle layer and reducing the time delay of the whole system.
(2) According to the invention, by embedding federal learning into edge calculation, a reliable cooperation mechanism is established, a centralized manager is saved, roadside units exist in a parameter server mode, and the risk of data leakage caused by a third-party server is avoided. And finally, returning corresponding optimal paths to the requests of different vehicles in real time through a federal learning training model, and simultaneously ensuring the privacy of original data and improving the data privacy protection degree.
Further, in step S2, the local history data DiSet of timestamps t comprising vehicle traveli={ti,1,ti.2,...,ti,jV set of historical speeds of vehiclei={vi,1,vi,2,...,vi,j}, historical position set loc of vehiclei={loci,1,loci,2,...,loci,jAnd the set of driving directions of the vehicle diri={diri,1,diri.2,...,diri,j}; wherein, ti,nRepresenting a sequence of time points at which data is collected by the vehicle, vi,nIndicating the speed of the vehicle at the moment of each data acquisition, loci,nIndicating the position coordinates of the vehicle at the moment of each data acquisition, diri,nJ, j represents the jth federal training data.
Further, in step S3, the aggregation model MkThe calculation formula of (2) is as follows:
where N represents the total number of vehicles participating in federal learning, DiRepresenting local history of the vehicle, miRepresenting a local model of a participating federally learned vehicle.
The beneficial effects of the further scheme are as follows: in the invention, a road traffic flow prediction model is constructed based on federal learning, and a plurality of vehicles are combined to participate in training the prediction model through federal learning, so that the privacy of data is protected.
Further, in step S4, the time limit TthThe calculation formula of (2) is as follows:
wherein v isiRepresenting the speed of travel, x, of vehicle ii,aInitial abscissa, y, indicating the beginning of connection of vehicle i into communication range of edge serveri,aInitial ordinate, x, indicating the beginning of a connection of a vehicle i into the communication range of an edge serveri,bInitial abscissa, y, representing the last connection of vehicle i out of the edge server communication rangei,bAn initial ordinate representing the last connection of vehicle i out of the edge server communication range.
Further, step S5 includes the following sub-steps:
s51: updating vehicles crOf real-time local State S'iTo roadside unit Rk;
S52: according to the updated vehicle crOf real-time local State S'iUsing roadside units RkThe edge server establishes a vehicle digital twin model DTiAnd vehicle digital twin network model Gk;
S53: by a plurality of wayside units RkUsing a vehicle digital twin network model GkAnd establishing a global digital twin model G (t) of the Internet of vehicles.
The beneficial effects of the further scheme are as follows: in the invention, by introducing the digital twin into the Internet of vehicles, a vehicle digital twin network model is provided, and analysis and calculation can be rapidly and effectively carried out on the edge layer formed by the roadside units. Vehicle digital twin model DTiIn effect, the information that the ith vehicle is within the current wayside unit service range, including that the vehicle's locality is for federal learningData set size and real-time information of the vehicle. Vehicle digital twin model DTiAre used to construct vertices in the digital twin network model.
Further, in step S51, the vehicle c is updatedrOf real-time local State S'iIs S'i={v′i,loc′i,t′i,diri'}, wherein, v'iIndicating updated vehicle crReal-time speed of, loc'iIndicating updated vehicle crReal-time position of t'iIndicating updated vehicle crReal time of (di)i' indicating updated vehicle crReal-time driving direction of the vehicle;
in step S52, vehicle digital twin model DTiIs given by the expression ofi={Mk,Di,s′iVehicle digital twin network model GkIs expressed as Gk=(Vk,Ek) Wherein, s'iRepresenting real-time information of the ith vehicle, MkRepresenting an aggregate model, DiIndicating a passing vehicle crLocal history data of Vk={DTi},VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, Ek={eij},EkAn edge indicating that the kth wayside unit forms a twin network according to the relationship between the vehicles, eijIndicating that when communication between two vehicles is possible, an edge in a twin network is established;
in step S53, the expression of the global digital twin model G (t) of the Internet of vehicles isWhere K represents the number of wayside units.
Further, in step S6, the roadside unit R is usedkReceiving vehicle crThe formula for periodically updating the global digital twin model of the Internet of vehicles is as follows:wherein G iskRepresenting a digital twin network model of a vehicle, VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, EkIndicating that the kth wayside unit forms an edge of the twin network according to the relationship between the vehicles,representing a hamiltonian operation.
The beneficial effects of the further scheme are as follows: in the invention, the global digital twin model of the Internet of vehicles is updated in real time, and the whole planned path is ensured to be up to date.
Further, step S7 includes the following sub-steps:
s71: by vehicles crTo roadside unit RkInitiating a way finding request reqr,i;
S72: by means of approaching vehicles crRoad side unit RkReceiving a way-finding request reqr,iAnd using roadside units RkThe edge server analyzes and obtains the optimal driving path p according to the global digital twin model G (t) of the Internet of vehiclesr,i(tr,dr) And a local prediction model M;
s73: communication via V2R, using roadside units RkThe polymerization model MkBack to vehicle cr;
S74: passing vehicle crSlave road side unit RkReceived aggregation model MkPredicting according to the real-time information of the vehicle to obtain an optimal path;
s75: repeating steps S71-S74 to update vehicle crThe optimal path planning of the intelligent traffic is completed by the optimal path and the local prediction model.
The beneficial effects of the further scheme are as follows: in the invention, a vehicle initiates a route searching request to a roadside unit, after receiving the route searching request, the adjacent roadside unit analyzes and calculates the optimal running route and the prediction model according to the local vehicle digital twin network model and returns the optimal running route and the prediction model to the requesting vehicle, and the vehicle obtains the optimal running route and updates the current optimal route in real time according to the obtained prediction model, thereby realizing the optimal control of the road flow. In step S73, the returned model should be a model obtained by aggregation at a roadside unit within the range of the current vehicle. The formula for its calculation is shown in claim 3. After the vehicles receive the aggregated model from the roadside unit, the aggregated model can be fed according to the real-time information of the current vehicles, and the optimal driving path at the current position can be obtained, so that the path planning of intelligent traffic is realized.
Further, in step S71, the way-finding request reqr,iIs expressed as reqr,i={vr,i,locr,i,tr,i,dirr,i,desr,iIn which v isr,iIndicating vehicle crCurrent running speed, locr,iIndicating vehicle crCurrent position, tr,iIndicating vehicle crTime of initiation of the request, dirr,iIndicating vehicle crCurrent direction of travel, desr,iIndicating vehicle crThe destination of (2);
in step S72, the optimal travel route pr,i(tr,dr) Is expressed as pr,i(tr,dr)=M(reqr,i) Wherein, trTime taken to represent a path, drIndicating the path distance, reqr,iRepresenting a way-finding request, and M representing a local prediction model.
Drawings
FIG. 1 is a flow chart of an intelligent traffic path planning method;
fig. 2 is a structural diagram of intelligent traffic path planning.
Detailed Description
The embodiments of the present invention will be further described with reference to the accompanying drawings.
As shown in fig. 1, the invention provides an intelligent traffic path planning method based on federal learning and digital twin, comprising the following steps:
s1: vehicle registration is carried out in a vehicle system, and vehicle identity information is verified;
s2: passing vehicle c according to verificationrLocal history data D ofiParticipating in federal learning and training local model mi;
S3: using roadside units RkEdge server of (2) aggregating all vehicles crLocal model m ofiTo obtain a polymerization model Mk;
S4: repeating the steps S2-S3 to judge the aggregation model MkWhether to converge to a predetermined accuracy thetathOr exceeding the time limit TthIf yes, go to step S5, otherwise return to step S2;
s5: updating vehicles crOf real-time local State S'iTo roadside unit RkAnd using a plurality of roadside units RkEstablishing a global digital twin model G (t) of the Internet of vehicles;
s6: using roadside units RkPeriodically updating the global digital twin model G (t) of the Internet of vehicles;
s7: according to the updated global digital twin model G (t) of the Internet of vehicles, utilizing the vehicle crTo roadside unit RkInitiating a way finding request reqr,iAnd updating the optimal path and the local prediction model in real time to complete the optimal path planning of the intelligent traffic.
In the embodiment of the present invention, as shown in fig. 1, in step S2, the local history data DiSet of timestamps t comprising vehicle traveli={ti,1,ti.2,...,ti,jV set of historical speeds of vehiclei={vi,1,vi,2,...,vi,j}, historical position set loc of vehiclei={loci,1,loci,2,...,loci,jAnd the set of driving directions of the vehicle diri={diri,1,diri.2,...,diri,j}; wherein, ti,nRepresenting a sequence of time points at which data is collected by the vehicle, vi,nIndicating the speed of the vehicle at the moment of each data acquisition, loci,nIndicating the position coordinates of the vehicle at the moment of each data acquisition, diri,nJ, j represents the jth federal training data.
In the practice of the inventionIn an example, as shown in FIG. 1, in step S3, the aggregation model MkThe calculation formula of (2) is as follows:
where N represents the total number of vehicles participating in federal learning, DiRepresenting local history of the vehicle, miRepresenting a local model of a participating federally learned vehicle.
In the invention, a road traffic flow prediction model is constructed based on federal learning, and a plurality of vehicles are combined to participate in training the prediction model through federal learning, so that the privacy of data is protected.
In the embodiment of the present invention, as shown in fig. 1, in step S4, the time limit TthThe calculation formula of (2) is as follows:
wherein v isiRepresenting the speed of travel, x, of vehicle ii,aInitial abscissa, y, indicating the beginning of connection of vehicle i into communication range of edge serveri,aInitial ordinate, x, indicating the beginning of a connection of a vehicle i into the communication range of an edge serveri,bInitial abscissa, y, representing the last connection of vehicle i out of the edge server communication rangei,bAn initial ordinate representing the last connection of vehicle i out of the edge server communication range.
In the embodiment of the present invention, as shown in fig. 1, step S5 includes the following sub-steps:
s51: updating vehicles crOf real-time local State S'iTo roadside unit Rk;
S52: according to the updated vehicle crOf real-time local State S'iUsing roadside units RkThe edge server establishes a vehicle digital twin model DTiAnd vehicle digital twin network model Gk;
S53: by a plurality of wayside units RkUsing a vehicle digital twin network model GkAnd establishing a global digital twin model G (t) of the Internet of vehicles.
In the invention, by introducing the digital twin into the Internet of vehicles, a vehicle digital twin network model is provided, and analysis and calculation can be rapidly and effectively carried out on the edge layer formed by the roadside units. Vehicle digital twin model DTiIn effect, is the information that the ith vehicle is within the current wayside unit service range, including the data set size for federal learning local to that vehicle and the vehicle's real-time information. Vehicle digital twin model DTiAre used to construct vertices in the digital twin network model.
In the embodiment of the invention, as shown in fig. 1, in step S51, the vehicle c is updatedrOf real-time local State S'iIs S'i={v′i,loc′i,t′i,diri'}, wherein, v'iIndicating updated vehicle crReal-time speed of, loc'iIndicating updated vehicle crReal-time position of t'iIndicating updated vehicle crReal time of (di)i' indicating updated vehicle crReal-time driving direction of the vehicle;
in step S52, vehicle digital twin model DTiIs given by the expression ofi={Mk,Di,s′iVehicle digital twin network model GkIs expressed as Gk=(Vk,Ek) Wherein, s'iRepresenting real-time information of the ith vehicle, MkRepresenting an aggregate model, DiIndicating a passing vehicle crLocal history data of Vk={DTi},VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, Ek={eij},EkAn edge indicating that the kth wayside unit forms a twin network according to the relationship between the vehicles, eijIndicating that when communication between two vehicles is possible, an edge in a twin network is established;
in step S53, Internet of vehicles GlobalThe expression of the digital twin model G (t) isWhere K represents the number of wayside units.
In the embodiment of the present invention, as shown in fig. 1, in step S6, a roadside unit R is usedkReceiving vehicle crThe formula for periodically updating the global digital twin model of the Internet of vehicles is as follows:wherein G iskRepresenting a digital twin network model of a vehicle, VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, EkIndicating that the kth wayside unit forms an edge of the twin network according to the relationship between the vehicles,representing Hamiltonian operation, wherein the Hamiltonian operation represents the change of a digital twin network formed by different wayside units, summing the changes of all the wayside units, and adding the summed changes to the global digital twin model at the current moment to form the global digital twin model at the next moment.
In the invention, the global digital twin model of the Internet of vehicles is updated in real time, and the whole planned path is ensured to be up to date.
In the embodiment of the present invention, as shown in fig. 1, step S7 includes the following sub-steps:
s71: by vehicles crTo roadside unit RkInitiating a way finding request reqr,i;
S72: by means of approaching vehicles crRoad side unit RkReceiving a way-finding request reqr,iAnd using roadside units RkThe edge server analyzes and obtains the optimal driving path p according to the global digital twin model G (t) of the Internet of vehiclesr,i(tr,dr) And a local prediction model M;
s73: communication via V2R, using roadside units RkThe polymerization model MkBack to vehicle cr;
S74: passing vehicle crSlave road side unit RkReceived aggregation model MkPredicting according to the real-time information of the vehicle to obtain an optimal path;
s75: repeating steps S71-S74 to update vehicle crThe optimal path planning of the intelligent traffic is completed by the optimal path and the local prediction model.
In the invention, a vehicle initiates a route searching request to a roadside unit, after receiving the route searching request, the adjacent roadside unit analyzes and calculates the optimal running route and the prediction model according to the local vehicle digital twin network model and returns the optimal running route and the prediction model to the requesting vehicle, and the vehicle obtains the optimal running route and updates the current optimal route in real time according to the obtained prediction model, thereby realizing the optimal control of the road flow. In step S73, the returned model should be a model obtained by aggregation at a roadside unit within the range of the current vehicle. The formula for its calculation is shown in claim 3. After the vehicles receive the aggregated model from the roadside unit, the aggregated model can be fed according to the real-time information of the current vehicles, and the optimal driving path at the current position can be obtained, so that the path planning of intelligent traffic is realized.
In the embodiment of the present invention, as shown in fig. 1, in step S71, the way-finding request reqr,iIs expressed as reqr,i={vr,i,locr,i,tr,i,dirr,i,desr,iIn which v isr,iIndicating vehicle crCurrent running speed, locr,iIndicating vehicle crCurrent position, tr,iIndicating vehicle crTime of initiation of the request, dirr,iIndicating vehicle crCurrent direction of travel, desr,iIndicating vehicle crThe destination of (2);
in step S72, the optimal travel route pr,i(tr,dr) Is expressed as pr,i(tr,dr)=M(reqr,i) Wherein, trTime taken to represent a path, drIndicating the path distance, reqr,iTo representAnd (4) a way finding request, wherein M represents a local prediction model.
The working principle and the process of the invention are as follows: the invention considers the following factors which influence the flow prediction and the path planning in the intelligent traffic system: firstly, the communication resources between the vehicle and the road infrastructure are limited, so that the communication of the vehicle has larger uncertainty; secondly, a large amount of data is transmitted and calculated and analyzed, and a large amount of bandwidth and calculation resources are consumed; in addition, the owner of the vehicle needs to protect the privacy journey of the historical data of the owner of the vehicle.
The car networking scene applied by the invention is as follows: on any road that covers the network. As shown in fig. 2, after vehicles running on a highway are registered in an intelligent traffic path planning system and verified, federal learning is performed locally, an edge server of a roadside unit aggregates the federal learning model to construct a vehicle digital twin model and a vehicle digital twin network model, a vehicle which needs to find a path initiates a request, and after calculation and analysis by the edge server, the roadside unit returns an optimal path.
The purpose of the invention is realized as follows: the invention relates to an intelligent traffic path planning method based on federal learning and digital twin technology, which is divided into two stages; the first stage is model construction: the method comprises the steps that a vehicle is registered in a system at first, after system verification, the vehicle participates in federal learning locally, all federal learning models are aggregated by an edge server of a roadside unit, after the models converge to preset precision, a vehicle digital twin model and a vehicle digital twin network model are established, and then a plurality of roadside units cooperatively establish a vehicle networking global digital twin model; the second phase is path planning: the vehicles send a road finding request to the roadside unit, the adjacent roadside unit analyzes and calculates the optimal running path and the prediction model according to the local vehicle digital twin network model after receiving the road finding request, and returns the optimal running path and the prediction model to the requesting vehicle, and the vehicles obtain the optimal running path and update the current optimal path in real time according to the obtained prediction model, so that the optimal control of the road flow is realized.
The invention has the beneficial effects that:
(1) the planning method is applied to the field of vehicle networking and aims to solve the problems that in the current road traffic system, the flow prediction and path planning are low in accuracy, high in time delay and prone to privacy disclosure. In the invention, in the process of traffic flow optimization control, by introducing a vehicle digital twin network model, the process of analyzing and calculating traffic flow data is effectively transferred to the roadside unit edge layer, thereby avoiding the risks brought by unreliable communication states and limited calculation resources of the vehicle layer and reducing the time delay of the whole system.
(2) According to the invention, by embedding federal learning into edge calculation, a reliable cooperation mechanism is established, a centralized manager is saved, roadside units exist in a parameter server mode, and the risk of data leakage caused by a third-party server is avoided. And finally, returning corresponding optimal paths to the requests of different vehicles in real time through a federal learning training model, and simultaneously ensuring the privacy of original data and improving the data privacy protection degree.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (6)
1. An intelligent traffic path planning method based on federal learning and digital twins is characterized by comprising the following steps:
s1: vehicle registration is carried out in a vehicle system, and vehicle identity information is verified;
s2: passing vehicle c according to verificationrLocal history data D ofiParticipating in federal learning and training local model mi;
S3: using roadside units RkEdge server of (2) aggregating all vehicles crLocal model m ofiTo obtain a polymerization model Mk;
In the step S3, the model M is aggregatedkThe calculation formula of (2) is as follows:
where N represents the total number of vehicles participating in federal learning, DiRepresenting local history of the vehicle, miA local model representing a vehicle participating in federal learning;
s4: repeating the steps S2-S3 to judge the aggregation model MkWhether to converge to a predetermined accuracy thetathOr exceeding the time limit TthIf yes, go to step S5, otherwise return to step S2;
s5: updating vehicles crOf real-time local State S'iTo roadside unit RkAnd using a plurality of roadside units RkEstablishing a global digital twin model G (t) of the Internet of vehicles;
the step S5 includes the following sub-steps:
s51: updating vehicles crOf real-time local State S'iTo roadside unit Rk;
S52: according to the updated vehicle crOf real-time local State S'iUsing roadside units RkThe edge server establishes a vehicle digital twin model DTiAnd vehicle digital twin network model Gk;
S53: by a plurality of wayside units RkUsing a vehicle digital twin network model GkEstablishing a global digital twin model G (t) of the Internet of vehicles;
in step S51, the vehicle c is updatedrOf real-time local State S'iIs S'i={v′i,loc′i,t′i,dir′iV 'therein'iIndicating updated vehicle crReal-time speed of, loc'iIndicating updated vehicle crReal-time position of t'iIndicating updated vehicle crReal time of dir'iIndicating updated vehicle crReal-time driving direction of the vehicle;
in the step S52, the vehicle digital twin model DTiIs given by the expression ofi={Mk,Di,S′SVehicle digital twin network model GkIs expressed as Gk=(Vk,Ek) Wherein, S'iIndicating updated vehicle crReal-time local state of (M)kRepresenting an aggregate model, DiIndicating a passing vehicle crLocal history data of Vk={DTi},VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, Ek={eij},EkAn edge indicating that the kth wayside unit forms a twin network according to the relationship between the vehicles, eijIndicating that when communication between two vehicles is possible, an edge in a twin network is established;
in the step S53, the expression of the global digital twin model g (t) of the internet of vehicles isWherein K represents the number of wayside units;
s6: using roadside units RkPeriodically updating the global digital twin model G (t) of the Internet of vehicles;
s7: according to the updated global digital twin model G (t) of the Internet of vehicles, utilizing the vehicle crTo roadside unit RkInitiating a way finding request reqr,iAnd updating the optimal path and the local prediction model in real time to complete the optimal path planning of the intelligent traffic.
2. The intelligent traffic path planning method based on federal learning and digital twin of claim 1, wherein in step S2, the local historical data DiSet of timestamps t comprising vehicle traveli={ti,1,ti.2,...,ti,jV set of historical speeds of vehiclei={vi,1,vi,2,...,vi,jCalendar of vehicleLocation history set loci={loci,1,loci,2,...,loci,jAnd the set of driving directions of the vehicle diri={diri,1,diri.2,...,diri,j}; wherein, ti,nRepresenting a sequence of time points at which data is collected by the vehicle, vi,nIndicating the speed of the vehicle at the moment of each data acquisition, loci,nIndicating the position coordinates of the vehicle at the moment of each data acquisition, diri,nJ, j represents the jth federal training data.
3. The intelligent traffic path planning method based on federal learning and digital twin of claim 1, wherein in step S4, the time limit T isthThe calculation formula of (2) is as follows:
wherein v isiRepresenting the speed of travel, x, of vehicle ii,aInitial abscissa, y, indicating the beginning of connection of vehicle i into communication range of edge serveri,aInitial ordinate, x, indicating the beginning of a connection of a vehicle i into the communication range of an edge serveri,bInitial abscissa, y, representing the last connection of vehicle i out of the edge server communication rangei,bAn initial ordinate representing the last connection of vehicle i out of the edge server communication range.
4. The intelligent traffic path planning method based on federal learning and digital twin of claim 1, wherein in step S6, a roadside unit R is usedkReceiving vehicle crThe formula for periodically updating the global digital twin model of the Internet of vehicles is as follows:wherein G iskRepresenting a digital twin network model of a vehicle, VkA vertex representing a twin network formed by all vehicles within the service range of the kth wayside unit, EkIndicating that the kth wayside unit forms an edge of the twin network according to the relationship between the vehicles,representing a hamiltonian operation.
5. The federal learning and digital twin-based intelligent transportation path planning method as claimed in claim 1, wherein the step S7 comprises the following sub-steps:
s71: by vehicles crTo roadside unit RkInitiating a way finding request reqr,i;
S72: by means of approaching vehicles crRoad side unit RkReceiving a way-finding request reqr,iAnd using roadside units RkThe edge server analyzes and obtains the optimal driving path p according to the global digital twin model G (t) of the Internet of vehiclesr,i(tr,dr) And a local prediction model M;
s73: communication via V2R, using roadside units RkThe polymerization model MkBack to vehicle cr;
S74: passing vehicle crSlave road side unit RkReceived aggregation model MkPredicting according to the real-time information of the vehicle to obtain an optimal path;
s75: repeating steps S71-S74 to update vehicle crThe optimal path planning of the intelligent traffic is completed by the optimal path and the local prediction model.
6. The intelligent traffic path planning method based on federal learning and digital twin as claimed in claim 5, wherein in step S71, the route searching request req is sentr,iIs expressed as reqr,i={vr,i,locr,i,tr,i,dirr,i,desr,iIn which v isr,iIndicating vehicle crCurrent running speed, locr,iIndicating vehiclescrCurrent position, tr,iIndicating vehicle crTime of initiation of the request, dirr,iIndicating vehicle crCurrent direction of travel, desr,iIndicating vehicle crThe destination of (2);
in step S72, the optimal travel route pr,i(tr,dr) Is expressed as pr,i(tr,dr)=M(reqr,i) Wherein, trTime taken to represent a path, drIndicating the path distance, reqr,iRepresenting a way-finding request, and M representing a local prediction model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011418422.2A CN112700639B (en) | 2020-12-07 | 2020-12-07 | Intelligent traffic path planning method based on federal learning and digital twins |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011418422.2A CN112700639B (en) | 2020-12-07 | 2020-12-07 | Intelligent traffic path planning method based on federal learning and digital twins |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112700639A CN112700639A (en) | 2021-04-23 |
CN112700639B true CN112700639B (en) | 2022-03-08 |
Family
ID=75506794
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011418422.2A Expired - Fee Related CN112700639B (en) | 2020-12-07 | 2020-12-07 | Intelligent traffic path planning method based on federal learning and digital twins |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112700639B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113177595B (en) * | 2021-04-29 | 2022-07-12 | 北京明朝万达科技股份有限公司 | Document classification model construction, training and testing method and model construction system |
CN113312177B (en) * | 2021-05-11 | 2024-03-26 | 南京航空航天大学 | Wireless edge computing system and optimizing method based on federal learning |
CN113177377A (en) * | 2021-05-24 | 2021-07-27 | 北京交通大学 | Intelligent urban rail transit network management system based on digital twins |
CN113313264B (en) * | 2021-06-02 | 2022-08-12 | 河南大学 | Efficient federal learning method in Internet of vehicles scene |
CN113283177B (en) * | 2021-06-16 | 2022-05-24 | 江南大学 | Mobile perception caching method based on asynchronous federated learning |
CN113419857B (en) * | 2021-06-24 | 2023-03-24 | 广东工业大学 | Federal learning method and system based on edge digital twin association |
CN113537514B (en) * | 2021-07-27 | 2023-07-25 | 北京邮电大学 | Digital twinning-based federal learning framework with high energy efficiency |
CN113758495B (en) * | 2021-09-10 | 2023-11-14 | 上海伯镭智能科技有限公司 | Unmanned road searching method based on mine |
CN113847917B (en) * | 2021-09-16 | 2024-01-16 | 西安电子科技大学 | Vehicle path planning method based on digital twin and user personalized requirements |
CN113867354B (en) * | 2021-10-11 | 2023-05-02 | 电子科技大学 | Regional traffic flow guiding method for intelligent cooperation of automatic driving multiple vehicles |
CN114266208B (en) * | 2022-03-03 | 2022-05-24 | 蘑菇物联技术(深圳)有限公司 | Methods, apparatus, media and systems for implementing dynamic prediction of piping pressure drop |
CN114863699B (en) * | 2022-06-14 | 2023-05-23 | 电子科技大学 | Urban vehicle-road cooperative traffic flow prediction method based on digital twin |
CN115499467B (en) * | 2022-09-06 | 2023-07-18 | 苏州大学 | Intelligent network vehicle connection test platform based on digital twinning and building method and system thereof |
CN116306323B (en) * | 2023-05-19 | 2023-08-08 | 中南大学 | Determination method and device of digital twin model, terminal equipment and medium |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104637328A (en) * | 2015-01-07 | 2015-05-20 | 浙江大学 | RSU (Roadside Unit)-based distributed real-time navigation method in vehicular ad hoc network |
CN106548646A (en) * | 2016-11-08 | 2017-03-29 | 西安电子科技大学宁波信息技术研究院 | Road information service system and method when being blocked up based on the city that mist is calculated |
CN108039053A (en) * | 2017-11-29 | 2018-05-15 | 南京锦和佳鑫信息科技有限公司 | A kind of intelligent network joins traffic system |
CN108291811A (en) * | 2015-11-26 | 2018-07-17 | 华为技术有限公司 | Switch the method and apparatus of trackside navigation elements in navigation system |
CN109000668A (en) * | 2018-05-25 | 2018-12-14 | 上海汽车集团股份有限公司 | Real-time intelligent air navigation aid based on car networking |
CN110940347A (en) * | 2018-09-21 | 2020-03-31 | 阿里巴巴集团控股有限公司 | Auxiliary vehicle navigation method and system |
CN111445720A (en) * | 2020-04-15 | 2020-07-24 | 中国电子科技集团公司第三十八研究所 | Indoor parking method and parking system based on digital twinning |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107389076B (en) * | 2017-07-01 | 2020-08-21 | 兰州交通大学 | Energy-saving real-time dynamic path planning method suitable for intelligent networked automobile |
WO2019156955A1 (en) * | 2018-02-06 | 2019-08-15 | Cavh Llc | Connected automated vehicle highway systems and methods for shared mobility |
US11373122B2 (en) * | 2018-07-10 | 2022-06-28 | Cavh Llc | Fixed-route service system for CAVH systems |
CN109141422A (en) * | 2018-07-24 | 2019-01-04 | 苏州溥诺斯智能科技有限公司 | A kind of vehicle positioning method and system based on roadside unit machine learning |
CN110930747B (en) * | 2018-09-20 | 2021-11-19 | 上海丰豹商务咨询有限公司 | Intelligent internet traffic service system based on cloud computing technology |
US11380145B2 (en) * | 2019-02-14 | 2022-07-05 | Oshkosh Corporation | Systems and methods for a virtual refuse vehicle |
US11620907B2 (en) * | 2019-04-29 | 2023-04-04 | Qualcomm Incorporated | Method and apparatus for vehicle maneuver planning and messaging |
CN111695734A (en) * | 2020-06-12 | 2020-09-22 | 中国科学院重庆绿色智能技术研究院 | Multi-process planning comprehensive evaluation system and method based on digital twin and deep learning |
-
2020
- 2020-12-07 CN CN202011418422.2A patent/CN112700639B/en not_active Expired - Fee Related
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104637328A (en) * | 2015-01-07 | 2015-05-20 | 浙江大学 | RSU (Roadside Unit)-based distributed real-time navigation method in vehicular ad hoc network |
CN108291811A (en) * | 2015-11-26 | 2018-07-17 | 华为技术有限公司 | Switch the method and apparatus of trackside navigation elements in navigation system |
CN106548646A (en) * | 2016-11-08 | 2017-03-29 | 西安电子科技大学宁波信息技术研究院 | Road information service system and method when being blocked up based on the city that mist is calculated |
CN108039053A (en) * | 2017-11-29 | 2018-05-15 | 南京锦和佳鑫信息科技有限公司 | A kind of intelligent network joins traffic system |
CN109000668A (en) * | 2018-05-25 | 2018-12-14 | 上海汽车集团股份有限公司 | Real-time intelligent air navigation aid based on car networking |
CN110940347A (en) * | 2018-09-21 | 2020-03-31 | 阿里巴巴集团控股有限公司 | Auxiliary vehicle navigation method and system |
CN111445720A (en) * | 2020-04-15 | 2020-07-24 | 中国电子科技集团公司第三十八研究所 | Indoor parking method and parking system based on digital twinning |
Non-Patent Citations (2)
Title |
---|
Federated Learning for Data Privacy Preservation in Vehicular Cyber-Physical Systems;卢云龙等;《IEEE Network》;20200602;全文 * |
Low-Latency Federated Learning and Blockchain for Edge Association in Digital Twin Empowered 6G Networks;卢云龙等;《IEEE Transactions on Industrial Informatics》;20200818;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112700639A (en) | 2021-04-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112700639B (en) | Intelligent traffic path planning method based on federal learning and digital twins | |
Zhao et al. | A fuzzy logic-based intelligent multiattribute routing scheme for two-layered SDVNs | |
Oubbati et al. | SEARCH: An SDN-enabled approach for vehicle path-planning | |
CN110361024A (en) | Utilize the dynamic lane grade automobile navigation of vehicle group mark | |
Wang et al. | Review of lane-changing maneuvers of connected and automated vehicles: models, algorithms and traffic impact analyses | |
Liu et al. | Parking-area-assisted spider-web routing protocol for emergency data in urban VANET | |
Noorani et al. | SDN-and fog computing-based switchable routing using path stability estimation for vehicular ad hoc networks | |
CN109612488B (en) | Big data micro-service-based mixed travel mode path planning system and method | |
Kim et al. | Navopt: Navigator assisted vehicular route optimizer | |
CN103236170B (en) | Novel dynamic-road-network-based traffic guidance method | |
CN111339554A (en) | User data privacy protection method based on mobile edge calculation | |
CN104080056A (en) | Message distribution method for vehicle-mounted self-organizing network based on connectivity probability perception | |
US20220306156A1 (en) | Route Planner and Decision-Making for Exploration of New Roads to Improve Map | |
Kamble et al. | On road intelligent vehicle path predication and clustering using machine learning approach | |
JP7321400B1 (en) | Estimation device and estimation method | |
Lv et al. | Edge computing task offloading for environmental perception of autonomous vehicles in 6G networks | |
Singh et al. | Leader election in cooperative adaptive cruise control based platooning | |
Yang et al. | Autonomic navigation system based on predicted traffic and VANETs | |
Vachan et al. | A user monitoring road traffic information collection using sumo and scheme for road surveillance with deep mind analytics and human behavior tracking | |
Zhu et al. | A prediction-based route guidance method toward intelligent and green transportation system | |
Li et al. | A traffic congestion aware vehicle-to-vehicle communication framework based on Voronoi diagram and information granularity | |
Liu et al. | Multiple UAVs collaborative traffic monitoring with intention-based communication | |
CN103906077A (en) | Road side unit placement method based on affinity propagation algorithm | |
Kumar et al. | Real-time data sharing, path planning and route optimization in urban traffic management | |
CN116403402A (en) | Traffic state prediction method for urban intersection area in network environment |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220308 |