CN114743379A - Beidou-based urban large-area road network traffic sensing method and system and cloud platform - Google Patents

Beidou-based urban large-area road network traffic sensing method and system and cloud platform Download PDF

Info

Publication number
CN114743379A
CN114743379A CN202210662467.7A CN202210662467A CN114743379A CN 114743379 A CN114743379 A CN 114743379A CN 202210662467 A CN202210662467 A CN 202210662467A CN 114743379 A CN114743379 A CN 114743379A
Authority
CN
China
Prior art keywords
road traffic
road
attribute
traffic
target
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
Application number
CN202210662467.7A
Other languages
Chinese (zh)
Other versions
CN114743379B (en
Inventor
邓维爱
李华栈
袁泽宇
彭文斌
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong Bangsheng Beidou Technology Co ltd
Original Assignee
Guangdong Bangsheng Beidou Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong Bangsheng Beidou Technology Co ltd filed Critical Guangdong Bangsheng Beidou Technology Co ltd
Priority to CN202210662467.7A priority Critical patent/CN114743379B/en
Publication of CN114743379A publication Critical patent/CN114743379A/en
Application granted granted Critical
Publication of CN114743379B publication Critical patent/CN114743379B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
    • G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
    • G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Abstract

The invention discloses a Beidou-based urban large-area road network traffic sensing method, a Beidou-based urban large-area road network traffic sensing system and a cloud platform, wherein the method comprises the following steps: firstly, acquiring an initial road traffic state to be sensed; then based on a first urban road network traffic perception optimization model, acquiring a road traffic attribute linear representation corresponding to initial road traffic state attributes, outputting at least two road traffic state attributes to be compared corresponding to the initial road traffic state attributes, and optimizing the initial road traffic state attributes into first confidence degrees of the corresponding road traffic state attributes to be compared, so as to determine target road traffic state attributes corresponding to the initial road traffic state attributes from the road traffic state attributes to be compared; and finally, carrying out standardized output on the attributes of the contrast road traffic states of all the targets to obtain the optimized target road traffic state of the initial road traffic state.

Description

Beidou-based urban large-area road network traffic sensing method and system and cloud platform
Technical Field
The invention relates to the technical field of Beidou, in particular to a Beidou-based urban large-area road network traffic sensing method, a Beidou-based urban large-area road network traffic sensing system and a cloud platform.
Background
At present, in each large city, due to the existence of scenes such as high-rise building dense areas, overpass areas and the like, the accurate perception of the urban road network cannot be guaranteed only by the existing Global Positioning System (GPS) or Beidou System, which not only affects the navigation experience of individual users, but also causes traffic management departments to be unable to accurately perceive the traffic conditions of the urban road network, and is unable to reasonably perform traffic control and scheduling.
Disclosure of Invention
The invention aims to provide a Beidou-based urban large-area road network traffic sensing method, a Beidou-based urban large-area road network traffic sensing system and a cloud platform.
In a first aspect, an embodiment of the present invention provides an urban large area road network traffic sensing method based on beidou, including:
acquiring an initial road traffic state to be sensed, wherein the initial road traffic state comprises at least one initial road traffic state attribute;
obtaining a linear representation of road traffic attributes corresponding to at least one initial road traffic state attribute in initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, each virtual Beidou road traffic data set comprises a second road traffic parameter and a first road traffic parameter obtained by optimizing the second road traffic parameter by the second urban road network traffic perception optimization model, the second urban road network traffic perception optimization model replaces weights of preset second urban road network traffic perception optimization models according to a second actual Beidou road traffic data set to construct, one second actual road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters, and the weight of the preset second urban road network traffic perception optimization model is determined according to the optimization results of the preset second urban road network traffic perception optimization model on the target second road traffic parameters and second loss function results among the auxiliary first road traffic parameters corresponding to the target second road traffic parameters;
outputting at least two road traffic state attributes to be contrasted corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized to the corresponding road traffic state attribute to be contrasted based on a first urban road network traffic perception optimization model according to road traffic attribute linear representation;
determining a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from road traffic state attributes to be compared according to a first confidence coefficient based on a first urban road network traffic perception optimization model;
and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state.
In a possible implementation manner, based on a first city road network traffic perception optimization model, according to a road traffic attribute linear representation, outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute, and a first confidence degree that each initial road traffic state attribute is optimized to a corresponding road traffic state attribute to be compared, includes:
performing benchmarking on the road traffic attribute linear representation into a second road traffic attribute linear space based on the incidence corresponding relation between the first road traffic attribute linear space of the initial perception strategy and the second road traffic attribute linear space of the optimized perception strategy in the first urban road network traffic perception optimization model to obtain a reference road traffic attribute linear representation;
according to the road traffic attribute linear representation of the undetermined road traffic state attribute value of the optimized perception strategy in the second road traffic attribute linear space, and the space distance linearly represented by the reference road traffic attribute, determining at least two road traffic state attributes to be detected corresponding to each initial road traffic state attribute in the road traffic state attribute value to be detected, and the first confidence degree of each initial road traffic state attribute optimized to the corresponding road traffic state attribute to be detected.
In a possible implementation, before acquiring the initial road traffic state to be sensed, the method further includes:
acquiring a second road traffic parameter and a plurality of first actual Beidou road traffic data sets, wherein each first actual Beidou road traffic data set comprises a configured first road traffic parameter and a configured second road traffic parameter which are matched with each other;
optimizing second road traffic parameters based on a second urban road network traffic perception optimization model to obtain optimized first road traffic parameters, and forming a virtual Beidou road traffic data set by the optimized first road traffic parameters and the corresponding second road traffic parameters;
determining a virtual Beidou road traffic data set and a first actual Beidou road traffic data set as actual Beidou road traffic data set training samples of a first urban road network traffic perception optimization model, wherein configured first road traffic parameters and optimized first road traffic parameters are first road traffic parameter samples, and configured second road traffic parameters and optimized second road traffic parameters are second road traffic parameter samples;
optimizing the first road traffic parameter sample based on a first urban road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimization perception strategy on the first urban road network traffic perception optimization model according to a first loss function result between the predicted second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample.
In a possible implementation manner, before optimizing the second road traffic parameter based on the second city road network traffic perception optimization model to obtain the optimized first road traffic parameter, the method further includes:
acquiring a second actual Beidou road traffic data set;
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in target second road traffic parameters based on a preset second urban road network traffic perception optimization model;
outputting at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute and a second confidence coefficient of each target road traffic state attribute optimized to the corresponding candidate initial road traffic attribute value according to the road traffic attribute linear representation corresponding to the target road traffic state attribute;
determining a target initial road traffic attribute value corresponding to each target road traffic state attribute from the candidate initial road traffic attribute values according to the second confidence coefficient;
carrying out standardized output on each target initial road traffic attribute value to obtain a predicted first road traffic parameter after optimization of a target second road traffic parameter;
calculating a second loss function result between the predicted first road traffic parameter of the same target second road traffic parameter and each corresponding auxiliary first road traffic parameter;
and replacing the weight of the preset second urban road network traffic perception optimization model according to the second loss function result to obtain a second urban road network traffic perception optimization model.
In one possible embodiment, outputting at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute and a second confidence degree that each target road traffic state attribute is optimized to the corresponding candidate initial road traffic attribute value according to the linear representation of the road traffic attribute corresponding to the target road traffic state attribute comprises:
acquiring an initial road traffic attribute value set, wherein the initial road traffic attribute value set comprises a plurality of initial road traffic attribute values;
performing label alignment on the road traffic attribute linear representation of the target road traffic state attribute into the first road traffic attribute linear space based on the incidence corresponding relation between the second road traffic attribute linear space of the optimized perception strategy and the first road traffic attribute linear space of the initial perception strategy in a preset second urban road network traffic perception optimization model to obtain a first reference road traffic attribute linear representation;
according to the linear representation of the road traffic attribute of the initial road traffic attribute value in the linear space of the first road traffic attribute and the space distance linearly represented by the first reference road traffic attribute, at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute in the initial road traffic attribute values are determined, and each target road traffic state attribute is optimized to be the second confidence coefficient of the corresponding candidate initial road traffic attribute value.
In a possible implementation manner, calculating a second loss function result between the predicted first road traffic parameter of the same target second road traffic parameter and each corresponding auxiliary first road traffic parameter includes:
for the same target second road traffic parameter, calculating a candidate initial road traffic attribute value corresponding to the target road traffic state attribute of the same target second road traffic parameter, and calculating a first matching degree between the target road traffic attribute values corresponding to the target road traffic state attribute in the auxiliary first road traffic parameter corresponding to the target second road traffic parameter;
calculating road traffic attribute loss between candidate initial road traffic attribute values corresponding to the target road traffic state attributes in the target second road traffic parameters and the target road traffic attribute values according to the first matching degree and the second confidence degree;
and carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a second loss function result between the predicted first road traffic parameter of the target second road traffic parameter and each corresponding auxiliary first road traffic parameter.
In a possible implementation manner, before optimizing the first road traffic parameter sample based on the first city road network traffic perception optimization model and obtaining the optimized predicted second road traffic parameter, the method further includes:
acquiring a third practical Beidou road traffic data set, wherein one third practical Beidou road traffic data set comprises a target first road traffic parameter and a plurality of corresponding auxiliary second road traffic parameters, and the auxiliary second road traffic parameters are optimized road traffic perception results of the target first road traffic parameters;
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in target first road traffic parameters based on a preset first urban road network traffic perception optimization model;
acquiring an optimized road traffic attribute value set, wherein the optimized road traffic attribute value set comprises a plurality of optimized road traffic attribute values;
performing label alignment on the road traffic attribute linear representation of the target road traffic state attribute into a second road traffic attribute linear space based on the incidence corresponding relation between a first road traffic attribute linear space of an initial perception strategy and a second road traffic attribute linear space of an optimized perception strategy in a preset first urban road network traffic perception optimization model to obtain a second reference road traffic attribute linear representation;
determining at least two candidate optimized road traffic attribute values corresponding to each target road traffic state attribute in the optimized road traffic attribute values and a third confidence coefficient of each target road traffic state attribute optimized to the corresponding candidate optimized road traffic attribute value according to the linear expression of the road traffic attribute of the optimized road traffic attribute values in the second road traffic attribute linear space and the space distance linearly expressed by the second reference road traffic attribute;
according to the third confidence coefficient, determining a target optimized road traffic attribute value corresponding to each target road traffic state attribute from the candidate optimized road traffic attribute values;
carrying out standardized output on each target optimized road traffic attribute value to obtain a predicted target second road traffic parameter after the target first road traffic parameter is optimized;
calculating a second matching degree between candidate optimized road traffic attribute values corresponding to the target road traffic state attribute of the same target first road traffic parameter and target road traffic attribute values corresponding to the target road traffic state attribute in auxiliary second road traffic parameters corresponding to the target first road traffic parameter;
calculating road traffic attribute loss between the candidate optimized road traffic attribute value corresponding to each target road traffic state attribute in the target first road traffic parameter and the target road traffic attribute value according to the second matching degree and the third confidence degree;
carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a third loss function result between a predicted target second road traffic parameter of the target first road traffic parameter and each corresponding auxiliary second road traffic parameter;
and replacing the weight of the preset first urban road network traffic perception optimization model according to the third loss function result to obtain the first urban road network traffic perception optimization model.
In one possible implementation, based on a first urban road network traffic perception optimization model, optimizing a first road traffic parameter sample to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimization perception strategy on the first urban road network traffic perception optimization model according to a first loss function result between the predicted second road traffic parameter and a second road traffic parameter sample corresponding to the first road traffic parameter sample, including:
acquiring road traffic attribute linear representation corresponding to each road traffic state attribute in a first road traffic parameter sample based on a first urban road network traffic perception optimization model;
according to the road traffic attribute linear representation, outputting at least two road traffic state attributes to be selected corresponding to each road traffic state attribute and a fourth confidence coefficient of each road traffic state attribute optimized to the corresponding road traffic state attribute to be selected;
according to the fourth confidence coefficient, determining a target road traffic state attribute corresponding to each road traffic state attribute from the road traffic state attributes to be selected, and carrying out standardized output on each target road traffic state attribute to obtain a predicted second road traffic parameter;
calculating and predicting a first loss function result between the second road traffic parameter and a second road traffic parameter sample corresponding to the first road traffic parameter sample;
and replacing the weight of the first urban road network traffic perception optimization model according to the first loss function result to obtain the trained first urban road network traffic perception optimization model.
In a second aspect, an embodiment of the present invention provides an urban large area road network traffic sensing system based on beidou, including:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring an initial road traffic state to be sensed, and the initial road traffic state comprises at least one initial road traffic state attribute; obtaining a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, each virtual Beidou road traffic data set comprises a second road traffic parameter and a first road traffic parameter obtained by optimizing the second road traffic parameter by the second urban road network traffic perception optimization model, the second urban road network traffic perception optimization model replaces a preset weight of the second urban road network traffic perception optimization model according to a second actual Beidou road traffic data set to construct, and one second actual road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters, and the weight of the preset second urban road network traffic perception optimization model is determined according to the optimization results of the preset second urban road network traffic perception optimization model on the target second road traffic parameters and second loss function results among the auxiliary first road traffic parameters corresponding to the target second road traffic parameters;
the output module is used for outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized into the corresponding road traffic state attribute to be compared based on the first urban road network traffic perception optimization model according to the linear representation of the road traffic attributes;
the optimization module is used for determining a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from road traffic state attributes to be compared based on a first city road network traffic perception optimization model and according to a first confidence coefficient; and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state.
In a third aspect, an embodiment of the present invention provides a Beidou based urban large area road network traffic awareness cloud platform, which includes a computer device for implementing at least one possible implementation manner in the first aspect, and a user terminal for receiving a target road traffic state.
Compared with the prior art, the beneficial effects provided by the invention comprise: the invention discloses a Beidou-based urban large-area road network traffic sensing method, a Beidou-based urban large-area road network traffic sensing system and a cloud platform, wherein an initial road traffic state to be sensed is obtained; then based on a first city road network traffic perception optimization model, acquiring a road traffic attribute linear representation corresponding to initial road traffic state attributes, outputting at least two road traffic state attributes to be contrasted corresponding to the initial road traffic state attributes, and optimizing the initial road traffic state attributes into a first confidence coefficient of the corresponding road traffic state attributes to be contrasted, so as to determine target contrast road traffic state attributes corresponding to the initial road traffic state attributes from the road traffic state attributes to be contrasted; and finally, carrying out standardized output on the attribute of each target contrast road surface traffic state to obtain the target road surface traffic state after the initial road surface traffic state is optimized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments will be briefly described below. It is appreciated that the following drawings depict only certain embodiments of the invention and are therefore not to be considered limiting of its scope. For a person skilled in the art, it is possible to derive other relevant figures from these figures without inventive effort.
Fig. 1 is a schematic structural diagram of interaction of a big dipper-based urban large-area road network traffic perception cloud platform according to an embodiment of the present invention;
fig. 2 is a schematic flow chart illustrating steps of a big dipper based urban large-area road network traffic perception method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an urban large-area road network traffic system based on the big dipper for implementing the urban large-area road network traffic sensing method based on the big dipper in fig. 2 according to the embodiment of the present invention;
fig. 4 is a schematic structural diagram of a computer device for executing the method for sensing traffic of the big dipper-based urban large-area road network in fig. 2 according to an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings.
Fig. 1 is an interaction schematic diagram of an urban large-area road network traffic perception cloud platform based on the Beidou satellite, which is provided by an embodiment of the disclosure. The Beidou based urban large-area road network traffic perception cloud platform can comprise a computer device 100 and a user terminal 200, wherein the user terminal 200 is connected with the computer device 100 based on a network, for example, based on a wireless network connection and the like.
The user terminal 200 may obtain an initial road traffic state to be sensed, where the initial road traffic state includes at least one attribute of the initial road traffic state, and then send the initial road traffic state to the computer device 100, so that the computer device 100 optimizes the initial road traffic state, and returns an optimization result to the user terminal 200, for example, send a target road traffic state after the initial road traffic state is optimized to the user terminal 200.
Computer apparatus 100 operable to: receiving an initial road traffic state sent by the user terminal 200; acquiring road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic state based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, the virtual Beidou road traffic data sets comprise second road traffic parameters, the second urban road network traffic perception optimization model is trained according to a second practical Beidou road traffic data set to obtain first road traffic parameters, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters; based on a first urban road network traffic perception optimization model, according to road traffic attribute linear representation, outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute, and a first confidence coefficient of each initial road traffic state attribute which is optimized to be the corresponding road traffic state attribute to be compared; determining a target comparison road surface traffic state attribute corresponding to each initial road surface traffic state attribute from road surface traffic state attributes to be compared according to a first confidence coefficient based on a first city road network traffic perception optimization model; based on a first urban road network traffic perception optimization model, carrying out standardized output on each target comparison road traffic state attribute to obtain a target road traffic state after the initial road traffic state is optimized; the target road traffic state is transmitted to the user terminal 200.
The embodiment of the application provides a Beidou-based urban large-area road network traffic perception method, relates to a traffic perception technology in the field of artificial intelligence, and particularly relates to machine optimization in the traffic perception technology. According to the embodiment of the application, the second city road network traffic perception optimization model can be trained according to a plurality of auxiliary first road traffic parameters, the diversity of the constructed virtual Beidou road traffic data set is increased, richer original information is provided for the training of the first city road network traffic perception optimization model, and the optimization quality is further improved.
The urban large-area road network traffic sensing method based on the Beidou can also relate to artificial intelligence cloud service in the technical field of cloud.
The Beidou-based urban large-area road network traffic sensing method can be applied to various optimized scenes, and can optimize the road traffic state determined based on the existing traffic sensing strategy to a more accurate road traffic state, wherein the optimized road traffic state determined based on the existing traffic sensing strategy and the more accurate sensing area type of the road traffic state are not limited, and the method can be applied to city centers, roundabout lines, communities and the like.
Referring to fig. 2, a specific flow of the urban large-area road network traffic sensing method based on the big dipper may be as follows:
101. and acquiring an initial road traffic state to be perceived, wherein the initial road traffic state comprises at least one initial road traffic state attribute.
The initial road traffic state belongs to an initial perception strategy, and a perception area of the initial perception strategy is not limited. The initial road traffic state is analyzed, and each initial road traffic state attribute in the initial road traffic state can be obtained.
Alternatively, in some embodiments, the initial road traffic state may be determined by a road traffic awareness application (e.g., existing navigation software) already installed in the corresponding terminal.
102. Acquiring a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, the virtual Beidou road traffic data sets comprise second road traffic parameters, and the second urban road network traffic perception optimization model is used for optimizing second road traffic parameters to obtain first road traffic parameters, the second urban road network traffic perception optimization model is obtained by training according to a second practical Beidou road traffic data set, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters.
Each initial road traffic state attribute in the initial road traffic state may be converted into a corresponding linear representation of the road traffic attribute based on a BERT model (Bidirectional encoderrepresentation from transforms) or the like.
The types of the first urban road Network traffic perception optimization model and the second urban road Network traffic perception optimization model may be Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), time recursive Neural Network (LSTM), bidirectional Recurrent Neural Network (BiRNN), and the like. It should be noted that the above examples should not be construed as limiting the first city road network traffic-aware optimization model and the second city road network traffic-aware optimization model.
The second urban road network traffic perception optimization model is obtained by training according to a second practical Beidou road traffic data set, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of corresponding auxiliary first road traffic parameters with the same perception target, the parameter types of the auxiliary first road traffic parameters are different, and the application environments of the auxiliary first road traffic parameters are different, namely, the second urban road network traffic perception optimization model is obtained by training according to a plurality of optimized road traffic perception results. Because the perception area of each optimized road traffic perception result is different from the application environment, the trained second urban road network traffic perception optimization model optimizes the second road traffic parameters, and can meet diversified perception requirements.
In this embodiment, the actual beidou road traffic data set includes the original road traffic parameters and the target road traffic parameters, the road traffic state determined based on the existing traffic perception strategy is the road traffic state that can be optimized to another road traffic state, and the road traffic state that is optimized to is a more accurate road traffic state. Each practical Beidou road traffic data set comprises mutually matched original road traffic parameters and target road traffic parameters. Road traffic conditions determined based on existing traffic awareness strategies and more accurate road traffic conditions are often distinct areas of awareness.
Urban road network traffic perception road surface traffic parameter in this embodiment, for the first urban road network traffic perception optimization model, the road surface traffic state determined based on the existing traffic perception strategy may be an initial perception strategy, and the more accurate road surface traffic state may be an optimized perception strategy. The actual Beidou road traffic data set can be specifically a data pair consisting of road traffic data belonging to an initial perception strategy and corresponding road traffic data belonging to an optimized perception strategy.
In some application scenarios, it is not easy to acquire a large number of actual Beidou road traffic data sets, especially in areas related to specific road perception scenarios, such as high buildings, overpasses, tunnels and the like, the actual Beidou road traffic data sets of the scenarios are difficult to acquire, and a plurality of initial perception results in the scenarios lack of corresponding optimized perception results.
Under the condition that lack actual big dipper road surface traffic data set, can construct virtual big dipper road surface traffic data set based on second city road network traffic perception optimization model, virtual big dipper road surface traffic data set includes first road surface traffic parameter and second road surface traffic parameter. Under the condition of obtaining the second road traffic parameters, the second road traffic parameters can be input into a second urban road network traffic perception optimization model, the first road traffic parameters matched with the second road traffic parameters are obtained, and then a virtual Beidou road traffic data set is constructed. Based on the virtual Beidou road traffic data set, the accuracy of perception optimization of some specific road perception scenes can be improved.
It should be noted that, an actual Beidou road traffic data set includes mutually matched original road traffic parameters and target road traffic parameters, where a relationship between the original road traffic parameters and the target road traffic parameters may be one-to-one, or one-to-many, and the like, which is not limited in this embodiment. That is, the original road traffic parameter may have one optimized road traffic sensing result or a plurality of optimized road traffic sensing results.
103. And outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized to the corresponding road traffic state attribute to be compared based on the first urban road network traffic perception optimization model according to the road traffic attribute linear representation.
Optionally, in some embodiments, the step "outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute according to a road traffic attribute linear representation based on a first city road network traffic perception optimization model, and optimizing each initial road traffic state attribute to a first confidence degree of the corresponding road traffic state attribute to be compared" may include:
performing benchmarking on the road traffic attribute linear representation to a second road traffic attribute linear space based on the association correspondence between the first road traffic attribute linear space of the initial perception strategy and the second road traffic attribute linear space of the optimized perception strategy in the first urban road network traffic perception optimization model to obtain a reference road traffic attribute linear representation;
according to the road traffic attribute linear representation of the undetermined road traffic state attribute values of the optimized perception strategy in the second road traffic attribute linear space and the space distance linearly represented by the reference road traffic attribute, determining at least two road traffic state attributes to be detected corresponding to each initial road traffic state attribute in the road traffic state attribute values to be detected, and the first confidence degree of each initial road traffic state attribute optimized to the corresponding road traffic state attribute to be detected.
The attribute value of the traffic state of the road surface to be determined belongs to an optimization perception strategy.
The road traffic attribute linear representation corresponding to the initial road traffic state attribute is mapped to a second road traffic attribute linear space to obtain a reference road traffic attribute linear representation, and specifically, the road traffic attribute linear representation of the initial road traffic state attribute can be subjected to convolution operation and pooling operation to obtain the reference road traffic attribute linear representation corresponding to the second road traffic attribute linear space.
The step "determining at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute in the road traffic state attribute values to be determined according to the linear representation of the road traffic attribute of the road traffic state attribute values to be determined in the second road traffic attribute linear space of the optimized perception strategy and the space distance linearly represented by the reference road traffic attribute, and optimizing each initial road traffic state attribute into the first confidence of the corresponding road traffic state attribute to be compared" may include:
calculating the road traffic attribute linear representation of the undetermined road traffic state attribute value of the optimized perception strategy in the second road traffic attribute linear space and the space distance linearly represented by the reference road traffic attribute;
determining at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute in the road traffic state attribute values to be determined according to the spatial distance;
and outputting a first confidence degree that each initial road traffic state attribute is optimized to the corresponding road traffic state attribute to be compared.
The spatial distance may be a euclidean distance, a cosine distance, a manhattan distance, and the like, which is not limited in this embodiment. The larger the spatial distance is, the larger the difference between the attribute value of the road traffic state to be determined and the perception target of the corresponding initial road traffic state attribute is, and the smaller the spatial distance is, the closer the attribute value of the road traffic state to be determined and the perception target of the corresponding initial road traffic state attribute are.
In this embodiment, the attribute value of the road traffic state to be determined, where the spatial distance is smaller than the preset distance, may be determined as the attribute of the road traffic state to be compared, where the preset distance may be set according to an actual situation, and this embodiment does not limit this. For example, the number of road traffic state attributes to be compared may be set according to the number of road traffic state attributes to be obtained.
And outputting a first confidence coefficient that each initial road traffic state attribute is optimized to a corresponding road traffic state attribute to be compared based on a full connection layer in the first urban road network traffic perception optimization model.
104. And determining a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from the road traffic state attributes to be compared according to the first confidence coefficient based on the first urban road network traffic perception optimization model.
For each initial road traffic state attribute, the to-be-determined contrast road traffic state attribute with the maximum first confidence coefficient may be used as the target contrast road traffic state attribute corresponding to the initial road traffic state attribute.
105. And based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state.
The first urban road network traffic perception optimization model and the second urban road network traffic perception optimization model are trained, and the second urban road network traffic perception optimization model is obtained by training according to a plurality of optimized road traffic perception results, so that the first road traffic parameters in the virtual Beidou road traffic data set constructed based on the second urban road network traffic perception optimization model are relatively more diverse; the first urban road network traffic perception optimization model is obtained by training a virtual Beidou road traffic data set constructed according to the second urban road network traffic perception optimization model, and the diversity of original perception results in the virtual Beidou road traffic data set can be increased for the first urban road network traffic perception optimization model, so that the optimization quality of the first urban road network traffic perception optimization model is enhanced based on richer original information.
The fusion mode of the target-to-road traffic state attributes may specifically be that the target-to-road traffic state attributes are spliced in a certain mode to obtain a target road traffic state.
It should be noted that the first urban road network traffic perception optimization model may be obtained by training a plurality of actual beidou road traffic data sets, where the actual beidou road traffic data set training samples may include a virtual beidou road traffic data set and a first actual beidou road traffic data set, where each first actual beidou road traffic data set includes a configured first road traffic parameter and a configured second road traffic parameter that are matched with each other, and the configured first road traffic parameter and the configured second road traffic parameter have the same perception target. The first city road network traffic perception optimization model can be provided for the optimization device according to the artificial intelligence after being trained by other equipment, or can be trained by the optimization device according to the artificial intelligence.
If the optimization device based on artificial intelligence is used for self-training, before the step of 'obtaining the initial road traffic state to be perceived', the Beidou-based urban large-area road network traffic perception method can further comprise the following steps of:
acquiring second road traffic parameters and a plurality of first actual Beidou road traffic data sets, wherein each first actual Beidou road traffic data set comprises configured first road traffic parameters and configured second road traffic parameters which are matched with each other;
optimizing the second road traffic parameters based on a second urban road network traffic perception optimization model to obtain optimized first road traffic parameters, and forming a virtual Beidou road traffic data set by the optimized first road traffic parameters and the corresponding second road traffic parameters;
determining a virtual Beidou road traffic data set and a first actual Beidou road traffic data set as actual Beidou road traffic data set training samples of a first urban road network traffic perception optimization model, wherein the configured first road traffic parameters and first road traffic parameters are first road traffic parameter samples, and the configured second road traffic parameters and second road traffic parameters are second road traffic parameter samples;
optimizing the first road traffic parameter sample based on a first urban road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimization perception strategy on the first urban road network traffic perception optimization model according to a first loss function result between the predicted second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample.
The practical Beidou road traffic data set training samples can be regarded as virtual Beidou road traffic data sets and first practical Beidou road traffic data sets and obtained.
The road traffic state determined by the first urban road network traffic perception optimization model based on the existing traffic perception strategy is the more accurate road traffic state of the second urban road network traffic perception optimization model, the more accurate road traffic state of the first urban road network traffic perception optimization model is the road traffic state determined by the second urban road network traffic perception optimization model based on the existing traffic perception strategy, namely, the road traffic state determined by the first urban road network traffic perception optimization model based on the existing traffic perception strategy is the initial perception strategy, and the more accurate road traffic state of the first urban road network traffic perception optimization model is the optimization perception strategy. It can be understood that the first city road network traffic perception optimization model can be regarded as a forward city road network traffic perception optimization model, and the second city road network traffic perception optimization model can be regarded as a reverse city road network traffic perception optimization model. The forward direction may be considered as a direction from the road traffic state determined based on the existing traffic perception strategy to a more precise road traffic state (i.e., from Left to Right (L2R)), and the reverse direction may be considered as a direction from the more precise road traffic state to the road traffic state determined based on the existing traffic perception strategy (i.e., from Right to Left (R2L)).
This application can optimize test road surface traffic parameter (being the second road surface traffic parameter that preceding embodiment mentioned) to demand road surface traffic parameter (being the first road surface traffic parameter that preceding embodiment mentioned) based on reverse city road network traffic perception optimization model, constructs virtual big dipper road surface traffic data set, and carries out data enhancement with actual big dipper road surface traffic data set, and then promotes the quality of forward city road network traffic perception optimization model, and its flow can include following step:
1001. training an inverse city road network traffic perception optimization model by using a diversity-driven training target (DDT), wherein diversity specifically refers to diversity of an optimized road traffic perception result; specifically, actual beidou road traffic data sets may be obtained, each actual beidou road traffic data set may include a target road traffic parameter (i.e., the target second road traffic parameter mentioned in the foregoing embodiment) and a plurality of original road traffic parameters (i.e., the auxiliary first road traffic parameter mentioned in the foregoing embodiment) corresponding thereto, the original road traffic parameters are optimized road traffic perception results of the target road traffic parameters, the target road traffic parameters are optimized based on the reverse city road network traffic perception optimization model to obtain optimized predicted first road traffic parameters, and the reverse city road network traffic perception optimization model is trained according to a loss function result between each optimized road traffic perception result corresponding to the target road traffic parameters and the predicted first road traffic parameters;
1002. optimizing test road traffic parameters by using a trained reverse city road network traffic perception optimization model, and outputting more diversified road traffic state display results determined based on the existing traffic perception strategy, wherein the road traffic state display results determined based on the existing traffic perception strategy are required road traffic parameters, and the required road traffic parameters and the corresponding test road traffic parameters form a virtual Beidou road traffic data set, so that the information of the road traffic state determined based on the existing traffic perception strategy in the constructed virtual Beidou road traffic data set is enriched;
1003. merging the virtual Beidou road traffic data set with an actual Beidou road traffic data set to obtain a merged actual Beidou road traffic data set, wherein the merged actual Beidou road traffic data set comprises merged original road traffic parameters and merged target road traffic parameters; and taking the combined original road traffic parameters as the input of the forward urban road network traffic perception optimization model to obtain optimized predicted second road traffic parameters, and training the forward urban road network traffic perception optimization model according to the cross entropy of Maximum Likelihood Estimation (MLE) between the predicted second road traffic parameters and corresponding target road traffic parameters.
Among them, it is emphasized that the second city road network traffic perception optimization model is already trained. The second urban road network traffic perception optimization model is obtained by training according to a plurality of optimized road traffic perception results, so that the first road traffic parameters in the virtual Beidou road traffic data set constructed based on the second urban road network traffic perception optimization model are relatively more diverse. For the first city road network traffic perception optimization model, the diversity of original perception results in the virtual Beidou road traffic data set can be increased, and original information is enriched.
Optionally, in some embodiments, the step "optimizing the first road traffic parameter sample based on the first city road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimization perception strategy on the first city road network traffic perception optimization model according to a first loss function result between the predicted second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample" may include:
acquiring road traffic attribute linear representation corresponding to each road traffic state attribute in a first road traffic parameter sample based on a first urban road network traffic perception optimization model;
according to the road traffic attribute linear representation, outputting at least two road traffic state attributes to be selected corresponding to each road traffic state attribute, and optimizing each road traffic state attribute into a fourth confidence coefficient of the corresponding road traffic state attribute to be selected;
according to the fourth confidence coefficient, determining a target road traffic state attribute corresponding to each road traffic state attribute from the road traffic state attributes to be selected, and carrying out standardized output on each target road traffic state attribute to obtain a predicted second road traffic parameter;
calculating and predicting a first loss function result between the second road traffic parameter and a second road traffic parameter sample corresponding to the first road traffic parameter sample;
and replacing the weight of the first urban road network traffic perception optimization model according to the first loss function result to obtain the trained first urban road network traffic perception optimization model.
In the optimization process of the first city road network traffic perception optimization model, the fourth confidence of the road traffic state attribute to be selected, which is close to the spatial distance in the optimized road traffic perception result (i.e. the corresponding second road traffic parameter sample), is increased.
The above embodiment may be referred to in the specific process of outputting the road traffic state attribute to be selected corresponding to each road traffic state attribute according to the road traffic attribute linear representation of each road traffic state attribute. The step of outputting at least two road traffic state attributes to be selected corresponding to each road traffic state attribute according to the linear expression of the road traffic attributes, and optimizing each road traffic state attribute to be a fourth confidence coefficient of the corresponding road traffic state attribute to be selected, may include:
performing benchmarking on the road traffic attribute linear representation to a second road traffic attribute linear space based on the association correspondence between the first road traffic attribute linear space of the initial perception strategy and the second road traffic attribute linear space of the optimized perception strategy in the first urban road network traffic perception optimization model to obtain a reference road traffic attribute linear representation;
according to the road traffic attribute linear representation of the attribute value of the road traffic state to be determined in the second road traffic attribute linear space of the optimized perception strategy and the space distance of the reference road traffic attribute linear representation, determining at least two road traffic state attributes to be selected corresponding to each road traffic state attribute in the attribute value of the road traffic state to be determined and the fourth confidence degree of each road traffic state attribute optimized to the corresponding road traffic state attribute to be selected.
For each road traffic state attribute, the road traffic state attribute to be selected with the highest fourth confidence coefficient may be used as the target road traffic state attribute corresponding to the road traffic state attribute.
When the first city road network traffic perception optimization model outputs each parameter in the perception result (namely predicting the second road traffic parameter), the first city road network traffic perception optimization model is not only guided by the optimized road traffic perception result (namely the corresponding second road traffic parameter sample) in the training set, but also considers parameters which are not covered by the reference parameter in the optimized road traffic perception result but can be correct.
The step of calculating and predicting a first loss function result between the second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample may include:
calculating and predicting road traffic attribute loss of each target road traffic state attribute in the second road traffic parameter and a corresponding sample road traffic state attribute in the second road traffic parameter sample;
and carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute in the predicted second road traffic parameters to obtain a first loss function result between the predicted second road traffic parameters and second road traffic parameter samples corresponding to the first road traffic parameter samples.
The road traffic attribute loss can be calculated according to the vector distance between the linear representation of the road traffic attribute corresponding to the target road traffic state attribute and the linear representation of the road traffic attribute corresponding to the sample road traffic state attribute. The vector distance may include a euclidean distance or a cosine distance, etc.
The fusion of the road traffic attribute losses corresponding to the target road traffic state attributes may specifically be performing weighted summation on the road traffic attribute losses corresponding to the target road traffic state attributes to obtain a first loss function result.
Optionally, in some embodiments, the step of "calculating and predicting a first loss function result between the second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample" may include:
obtaining confidence coefficient of the sample road traffic state attribute in the second road traffic parameter sample optimized by each road traffic state attribute;
and according to the confidence coefficient, calculating and predicting a first loss function result between the second road traffic parameter and a second road traffic parameter sample corresponding to the first road traffic parameter sample.
Optionally, in some embodiments, the step "replace a weight of the first city road network traffic perception optimization model according to a result of the first loss function to obtain a trained first city road network traffic perception optimization model", may specifically include: and determining the weight of the first urban road network traffic perception optimization model by adopting a back propagation algorithm (BP) or a stochastic gradient Descent algorithm (SGD), and optimizing the weight of the first urban road network traffic perception optimization model according to the first loss function result to enable the first loss function result to be smaller than a preset loss function result so as to obtain the trained first urban road network traffic perception optimization model. The result of the preset loss function can be set according to the actual situation.
The second city road network traffic perception optimization model is obtained by training according to a second practical Beidou road traffic data set, and one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of corresponding auxiliary first road traffic parameters. The second city road network traffic perception optimization model can be provided for the optimization device according to the artificial intelligence after being trained by other equipment, or can be trained by the optimization device according to the artificial intelligence.
Optionally, in this embodiment, the first urban road network traffic perception optimization model may be obtained by training in advance according to a third actual beidou road traffic data set. The third practical Beidou road traffic data set comprises a target first road traffic parameter and a plurality of corresponding auxiliary second road traffic parameters, the auxiliary second road traffic parameters are optimized road traffic perception results of the target first road traffic parameter, and the auxiliary second road traffic parameters are different in parameter type and different in application environment. The first city road network traffic perception optimization model can be provided for the optimization device according to the artificial intelligence after being trained by other equipment, or can be trained by the optimization device according to the artificial intelligence.
If the second urban road network traffic perception optimization model is trained by the optimization device according to artificial intelligence, before the step "optimizing the second road traffic parameters based on the second urban road network traffic perception optimization model to obtain the optimized first road traffic parameters", the urban large-area road network traffic perception method based on the big dipper may further include:
acquiring a second practical Beidou road traffic data set, wherein one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of corresponding auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameter;
optimizing the target second road traffic parameter based on a preset second urban road network traffic perception optimization model to obtain an optimized predicted first road traffic parameter;
calculating a second loss function result between the predicted first road traffic parameter of the same target second road traffic parameter and each corresponding auxiliary first road traffic parameter;
and replacing the weight of the preset second urban road network traffic perception optimization model according to the second loss function result to obtain a second urban road network traffic perception optimization model.
And optimizing the weight of the preset second city road network traffic perception optimization model according to a second loss function result, so that the second loss function result is smaller than the preset loss function result, and obtaining the second city road network traffic perception optimization model. The preset loss function result can be set according to actual conditions.
Optionally, in some embodiments, the step "optimizing the target second road traffic parameter based on a preset second city road network traffic perception optimization model to obtain an optimized predicted first road traffic parameter" may include:
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in the target second road traffic parameters based on a preset second urban road network traffic perception optimization model;
outputting at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute according to the road traffic attribute linear representation, and optimizing each target road traffic state attribute into a second confidence coefficient of the corresponding candidate initial road traffic attribute value;
determining a target initial road traffic attribute value corresponding to each target road traffic state attribute from the candidate initial road traffic attribute values according to the second confidence;
and carrying out standardized output on each target initial road traffic attribute value to obtain a predicted first road traffic parameter after the target second road traffic parameter is optimized.
For each target road traffic state attribute, the candidate initial road traffic attribute value with the largest second confidence coefficient may be used as the target initial road traffic attribute value corresponding to the target road traffic state attribute, and the target initial road traffic attribute value may also be used as the optimization result of the target road traffic state attribute. The fusion mode of the initial road traffic attribute values of each target may specifically be that the initial road traffic attribute values of each target are spliced according to a certain mode to obtain a predicted first road traffic parameter.
When the second urban road network traffic perception optimization model outputs each parameter in the perception result (namely, the first road traffic parameter is predicted), the second urban road network traffic perception optimization model is not only guided by the optimized road traffic perception result (namely, the corresponding auxiliary first road traffic parameter) in the training set, but also considers parameters which are not covered by the reference parameter in the optimized road traffic perception result but can be correct (similar to the perception target of the optimized road traffic perception result).
Optionally, in some embodiments, the step of outputting at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute according to the linear representation of the road traffic attributes, and optimizing each target road traffic state attribute to a second confidence degree of the corresponding candidate initial road traffic attribute value, may include:
acquiring an initial road traffic attribute value set, wherein the initial road traffic attribute value set comprises a plurality of initial road traffic attribute values;
performing label alignment on the road traffic attribute linear representation of the target road traffic state attribute into the first road traffic attribute linear space based on the incidence corresponding relation between the second road traffic attribute linear space of the optimized perception strategy and the first road traffic attribute linear space of the initial perception strategy in a preset second urban road network traffic perception optimization model to obtain a first reference road traffic attribute linear representation;
according to the linear representation of the road traffic attribute of the initial road traffic attribute value in the linear space of the first road traffic attribute and the space distance linearly represented by the first reference road traffic attribute, at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute in the initial road traffic attribute values are determined, and each target road traffic state attribute is optimized to be the second confidence coefficient of the corresponding candidate initial road traffic attribute value.
The road traffic attribute linear representation of the target road traffic state attribute is mapped into a first road traffic attribute linear space to obtain a first reference road traffic attribute linear representation, and specifically, the road traffic attribute linear representation of the target road traffic state attribute can be subjected to convolution operation and pooling operation to obtain a first reference road traffic attribute linear representation corresponding to the first road traffic attribute linear space.
The spatial distance may be a euclidean distance, a cosine distance, a manhattan distance, and the like, which is not limited in this embodiment. The initial road traffic attribute value with the spatial distance smaller than the preset distance may be determined as the candidate initial road traffic attribute value, and the preset distance may be set according to an actual situation, which is not limited in this embodiment.
In the optimization process of presetting the second urban road network traffic perception optimization model, a second confidence coefficient of a candidate initial road traffic attribute value close to the spatial distance in the optimized road traffic perception result (namely, the corresponding auxiliary first road traffic parameter) is increased.
Optionally, in some embodiments, the step of "calculating a second loss function result between the predicted first road traffic parameter of the same target second road traffic parameter and each corresponding auxiliary first road traffic parameter" may include:
calculating candidate initial road traffic attribute values corresponding to the target road traffic state attributes of the same target second road traffic parameters, and calculating a first matching degree between the target road traffic attribute values corresponding to the target road traffic state attributes in the auxiliary first road traffic parameters corresponding to the target second road traffic parameters;
calculating road traffic attribute loss between candidate initial road traffic attribute values corresponding to the target road traffic state attributes in the target second road traffic parameters and the target road traffic attribute values according to the first matching degree and the second confidence degree;
and carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a second loss function result between the predicted first road traffic parameter of the target second road traffic parameter and each corresponding auxiliary first road traffic parameter.
The target road traffic state attribute corresponds to a plurality of candidate initial road traffic attribute values, and the first matching degree comprises the matching degree between each candidate initial road traffic attribute value corresponding to the target road traffic state attribute and the corresponding target road traffic attribute value.
The first matching degree may be specifically measured based on a vector distance, which is specifically a vector distance between a linear representation of the road traffic property of the candidate initial road traffic property value and a linear representation of the road traffic property of the target road traffic property value. The larger the vector distance is, the smaller the first matching degree is; the smaller the vector distance, the greater the first degree of matching. The vector distance may be a euclidean distance, a cosine distance, or the like.
The road traffic attribute loss corresponding to each target road traffic state attribute is subjected to standardized output, and specifically, the road traffic attribute loss corresponding to each target road traffic state attribute is subjected to weighted summation.
And pre-training a traffic perception optimization model of the first urban road network.
If the first urban road network traffic perception optimization model is pre-trained by the optimization device according to artificial intelligence, before the step "optimizing the first road traffic parameter sample based on the first urban road network traffic perception optimization model to obtain the optimized predicted second road traffic parameter", the urban large area road network traffic perception method based on the big dipper may further include:
acquiring a third practical Beidou road traffic data set, wherein one third practical Beidou road traffic data set comprises a target first road traffic parameter and a plurality of corresponding auxiliary second road traffic parameters, and the auxiliary second road traffic parameters are optimized road traffic perception results of the target first road traffic parameter;
optimizing the target first road traffic parameters based on a preset first city road network traffic perception optimization model to obtain optimized predicted target second road traffic parameters;
calculating a third loss function result between a predicted target second road traffic parameter of the same target first road traffic parameter and each corresponding auxiliary second road traffic parameter;
and replacing the weight of the preset first urban road network traffic perception optimization model according to the third loss function result to obtain a first urban road network traffic perception optimization model.
And optimizing the weight of the preset first city road network traffic perception optimization model according to a third loss function result, so that the third loss function result is smaller than the preset loss function result, and obtaining the first city road network traffic perception optimization model. The preset loss function result can be set according to actual conditions.
Optionally, in some embodiments, the step "optimizing the target first road traffic parameter based on a preset first city road network traffic perception optimization model to obtain an optimized predicted target second road traffic parameter" may include:
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in target first road traffic parameters based on a preset first urban road network traffic perception optimization model;
outputting at least two candidate optimized road traffic attribute values corresponding to each target road traffic state attribute according to the road traffic attribute linear representation, and optimizing each target road traffic state attribute into a third confidence coefficient of the corresponding candidate optimized road traffic attribute value;
according to the third confidence coefficient, determining a target optimized road traffic attribute value corresponding to each target road traffic state attribute from the candidate optimized road traffic attribute values;
and carrying out standardized output on each target optimized road traffic attribute value to obtain a predicted target second road traffic parameter after the target first road traffic parameter is optimized.
For each target road traffic state attribute, the candidate optimized road traffic attribute value with the maximum third confidence coefficient may be used as the target optimized road traffic attribute value corresponding to the target road traffic state attribute, and the target optimized road traffic attribute value may also be used as the optimization result of the target road traffic state attribute. The fusion mode of the optimized road traffic attribute values of each target may specifically be to splice the optimized road traffic attribute values of each target in a certain mode to obtain a predicted target second road traffic parameter.
When each parameter in the sensing result (namely the predicted target second road traffic parameter) output by the first city road network traffic sensing optimization model is preset, the first city road network traffic sensing optimization model is guided by the optimization of the road traffic sensing result (namely the corresponding auxiliary second road traffic parameter) in the training set, and parameters which are not covered by the reference parameter in the optimized road traffic sensing result but can be correct (which are close to the sensing target of the optimized road traffic sensing result) can be considered.
Optionally, in some embodiments, the step "outputting at least two candidate optimized road traffic attribute values corresponding to each target road traffic state attribute according to the linear representation of the road traffic attributes, and the third confidence that each target road traffic state attribute is optimized to the corresponding candidate optimized road traffic attribute value" may include:
acquiring an optimized road traffic attribute value set, wherein the optimized road traffic attribute value set comprises a plurality of optimized road traffic attribute values;
performing label alignment on the road traffic attribute linear representation of the target road traffic state attribute into a second road traffic attribute linear space based on the incidence corresponding relation between a first road traffic attribute linear space of an initial perception strategy and a second road traffic attribute linear space of an optimized perception strategy in a preset first urban road network traffic perception optimization model to obtain a second reference road traffic attribute linear representation;
and determining at least two candidate optimized road traffic attribute values corresponding to each target road traffic state attribute in the optimized road traffic attribute values and a third confidence coefficient of each target road traffic state attribute optimized to the corresponding candidate optimized road traffic attribute value according to the linear expression of the road traffic attribute of the optimized road traffic attribute values in the second road traffic attribute linear space and the space distance of the linear expression of the second reference road traffic attribute.
The road traffic attribute linear representation of the target road traffic state attribute is mapped into a second road traffic attribute linear space to obtain a second reference road traffic attribute linear representation, and specifically, the road traffic attribute linear representation of the target road traffic state attribute can be subjected to convolution operation and pooling operation to obtain a second reference road traffic attribute linear representation corresponding to the second road traffic attribute linear space.
The spatial distance may be a euclidean distance, a cosine distance, a manhattan distance, and the like, which is not limited in this embodiment. The optimized road traffic attribute value with the spatial distance smaller than the preset distance may be determined as the candidate optimized road traffic attribute value, and the preset distance may be set according to an actual situation, which is not limited in this embodiment.
In the optimization process of presetting the first city road network traffic perception optimization model, a third confidence coefficient of a candidate optimized road traffic attribute value close to a spatial distance in an optimized road traffic perception result (namely, a corresponding auxiliary second road traffic parameter) is increased.
Optionally, in some embodiments, the step "calculating a third loss function result between the predicted target second road traffic parameter of the same target first road traffic parameter and each corresponding auxiliary second road traffic parameter" may include:
calculating a second matching degree between candidate optimized road traffic attribute values corresponding to the target road traffic state attribute of the same target first road traffic parameter and target road traffic attribute values corresponding to the target road traffic state attribute in an auxiliary second road traffic parameter corresponding to the target first road traffic parameter;
calculating road traffic attribute loss between the candidate optimized road traffic attribute value corresponding to each target road traffic state attribute in the target first road traffic parameter and the target road traffic attribute value according to the second matching degree and the third confidence degree;
and carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a third loss function result between the predicted target second road traffic parameter of the target first road traffic parameter and each corresponding auxiliary second road traffic parameter.
The target road traffic state attribute corresponds to a plurality of candidate optimized road traffic attribute values, and the second matching degree comprises the matching degree between each candidate optimized road traffic attribute value corresponding to the target road traffic state attribute and the corresponding target road traffic attribute value.
The second matching degree may be specifically measured based on a vector distance, which is specifically a vector distance between the linear representation of the road traffic property of the candidate optimized road traffic property value and the linear representation of the road traffic property of the target road traffic property value. The larger the vector distance is, the smaller the second matching degree is; the smaller the vector distance, the greater the second degree of matching. The vector distance may be a euclidean distance, a cosine distance, or the like.
The road traffic attribute loss corresponding to each target road traffic state attribute is subjected to standardized output, and specifically, the road traffic attribute loss corresponding to each target road traffic state attribute is subjected to weighted summation.
The specific calculation method of the third loss function result may refer to the description of the related embodiment (calculation of the second loss function result) in the training of the second city road network traffic perception optimization model, and is not described herein again.
As can be seen from the above, the present embodiment may acquire an initial road traffic state to be perceived, where the initial road traffic state includes at least one initial road traffic state attribute; acquiring a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, the virtual Beidou road traffic data sets comprise second road traffic parameters, the second urban road network traffic perception optimization model is trained according to a second practical Beidou road traffic data set to obtain first road traffic parameters, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters; based on a first urban road network traffic perception optimization model, according to road traffic attribute linear representation, outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute, and a first confidence coefficient of each initial road traffic state attribute which is optimized to be the corresponding road traffic state attribute to be compared; determining a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from road traffic state attributes to be compared according to a first confidence coefficient based on a first urban road network traffic perception optimization model; and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state. According to the embodiment of the application, the second city road network traffic perception optimization model can be trained according to a plurality of auxiliary first road traffic parameters, the diversity of the constructed virtual Beidou road traffic data set is increased, richer original information is provided for the training of the first city road network traffic perception optimization model, and the optimization quality is further improved.
According to the method described in the foregoing embodiment, the following will be described in further detail by way of example in which the optimization apparatus according to artificial intelligence is specifically integrated in a server, which may specifically be a cloud server or the like.
The embodiment of the application provides a Beidou-based urban large-area road network traffic sensing method, and the specific flow of the Beidou-based urban large-area road network traffic sensing method can be as follows:
201. the server receives an initial road traffic state to be sensed, which is sent by the terminal and comprises at least one initial road traffic state attribute.
202. The server obtains a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic state based on the first city road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, the virtual Beidou road traffic data sets comprise second road traffic parameters, and optimizing second road traffic parameters by a second urban road network traffic perception optimization model to obtain first road traffic parameters, wherein the second urban road network traffic perception optimization model is obtained by training according to a second practical Beidou road traffic data set, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters.
In this embodiment, the first city road network traffic perception optimization model and the second city road network traffic perception optimization model may be any Neural network Machine optimization (NMT) model. For example, NMT models based on Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), and Self-Attention (Self-Attention), NMT models using RNN, CNN, and Self-Attention in a mixed manner, and the like may be used.
The neural network machine optimization can use an encoder to represent the original traffic control road condition data into a vector sequence, and then a decoder generates a subsequent sensing result according to the original vector representation and the output sensing result information.
The second urban road network traffic perception optimization model is obtained by training according to a second practical Beidou road traffic data set, and one second practical Beidou road traffic data set comprises one target second road traffic parameter and a plurality of corresponding auxiliary first road traffic parameters, namely the second urban road network traffic perception optimization model is obtained by training according to a plurality of optimized road traffic perception results. Because the perception area of each optimized road traffic perception result is different from the application environment, the trained second urban road network traffic perception optimization model optimizes the second road traffic parameters, and traffic road perception with more scenes and demands can be achieved.
203. The server outputs at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized to the corresponding road traffic state attribute to be compared based on the first urban road network traffic perception optimization model and according to the road traffic attribute linear representation.
204. And the server determines a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from the road traffic state attributes to be compared according to the first confidence coefficient based on the first urban road network traffic perception optimization model.
205. And the server carries out standardized output on the attributes of the comparison road traffic states of the targets based on the first urban road network traffic perception optimization model to obtain the optimized target road traffic state of the initial road traffic state.
206. And the server sends the target road traffic state to the terminal.
The first urban road network traffic perception optimization model and the second urban road network traffic perception optimization model are trained, and the second urban road network traffic perception optimization model is obtained by training according to a plurality of optimized road traffic perception results, so that first road traffic parameters in a virtual Beidou road traffic data set constructed based on the second urban road network traffic perception optimization model are more diverse; the first urban road network traffic perception optimization model is obtained by training a virtual Beidou road traffic data set constructed according to the second urban road network traffic perception optimization model, and the diversity of original perception results in the virtual Beidou road traffic data set can be increased for the first urban road network traffic perception optimization model, so that the optimization quality of the first urban road network traffic perception optimization model is enhanced based on richer original information.
After receiving the target road traffic state, the terminal can display the optimized target road traffic state on a display of the electronic device.
In a specific embodiment, a first city road network traffic perception optimization model is used as a forward city road network traffic perception optimization model, a second city road network traffic perception optimization model is used as a reverse city road network traffic perception optimization model, the forward city road network traffic perception optimization model optimizes an original road traffic state to a target road traffic state, the reverse city road network traffic perception optimization model optimizes the target road traffic state to the original road traffic state, and specific training processes of the forward city road network traffic perception optimization model and the reverse city road network traffic perception optimization model are as follows:
2001. acquiring road traffic state training data (which can be considered as a first actual Beidou road traffic data set mentioned in the previous embodiment), wherein the road traffic state training data specifically comprises an original road traffic state A and a target road traffic state B;
2002. selecting a training direction, wherein a scheme for increasing the diversity of sensing results of the forward urban road network traffic sensing optimization model can be selected, and a scheme for improving the optimization quality of the forward urban road network traffic sensing optimization model based on an improved reverse optimization technology can also be selected;
2003. if the scheme for increasing the diversity of the sensing results of the forward urban road network traffic sensing optimization model is selected in step 2002, the forward urban road network traffic sensing optimization model can be trained according to a diversity-driven training target (DDT), specifically, various target end candidate sensing results can be obtained, both the target end candidate sensing results and the target road traffic state B can be used as more accurate road traffic state optimization road traffic sensing results (which can be considered as auxiliary second road traffic parameters mentioned in the foregoing embodiment) of the original road traffic state a, the road traffic sensing results and the original road traffic state a are optimized according to the more accurate road traffic states, and the forward urban road network traffic sensing optimization model is trained;
2004. according to the training in the step 2003, adjusting the weight of the forward urban road network traffic perception optimization model so that a loss function result between the optimized predicted target second road traffic parameter and the more accurate road traffic state optimization road traffic perception result meets a preset condition, and obtaining a trained forward urban road network traffic perception optimization model M which can increase the diversity of perception results;
2005. if the scheme for improving the optimization quality of the forward urban road network traffic perception optimization model based on the improved reverse optimization technology is selected in step 2002, the reverse urban road network traffic perception optimization model can be trained according to a diversity-driven training target, specifically, various original candidate perception results can be obtained, the original candidate perception results and the original road traffic state a can be used as the road traffic state optimization road traffic perception results (which can be regarded as the auxiliary first road traffic parameters mentioned in the foregoing embodiment) determined based on the existing traffic perception strategy of the target road traffic state B, and the reverse urban road network traffic perception optimization model is trained according to the road traffic state optimization road perception results determined based on the existing traffic perception strategy and the target road traffic state B;
2006. according to the training of the step 2005, adjusting the weight of the reverse city road network traffic perception optimization model so that a loss function result between the optimized predicted first road traffic parameter and the road traffic state optimized road traffic perception result determined based on the existing traffic perception strategy meets a preset condition, and obtaining the trained reverse city road network traffic perception optimization model;
2007. acquiring simulated target road traffic data C;
2008. optimizing the simulated target road traffic data C according to the reverse city road network traffic perception optimization model trained in the step 2006 to obtain a pseudo-original road traffic state D, wherein the pseudo-original road traffic state D and the simulated target road traffic data C form a virtual Beidou road traffic data set;
2009. merging the virtual Beidou road traffic data set and the first actual Beidou road traffic data set to obtain an actual Beidou road traffic data set training sample of the forward urban road network traffic perception optimization model, wherein the actual Beidou road traffic data set training sample comprises an original road traffic state (A + D) and a target road traffic state (B + C);
2010. the method comprises the steps that an original road traffic state (A + D) can be used as input of a forward urban road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and optimization training from a road traffic state determined based on the existing traffic perception strategy to a more accurate road traffic state is carried out on the forward urban road network traffic perception optimization model according to a cross entropy of maximum likelihood estimation between the predicted second road traffic parameter and a corresponding target road traffic state (B + C);
2011. based on the training in the step 2010, adjusting the weight of the forward urban road network traffic perception optimization model so that the cross entropy of the maximum likelihood estimation meets the preset condition, and obtaining a trained forward urban road network traffic perception optimization model M1; because reverse city road network traffic perception optimization model possesses the various perception results of stronger output ability, so the actual big dipper road surface traffic data set training sample after the combination has contained more traffic road surface perception information, can promote the optimization quality of final forward city road network traffic perception optimization model M1.
Optionally, in some embodiments, training for increasing the diversity of the perception results may be performed only on the forward urban road network traffic perception optimization model (see steps 2001-2004), or the optimization quality of the forward urban road network traffic perception optimization model may be enhanced only on the basis of the improved reverse optimization technology (see steps 2001, 2002, and steps 2005-2011).
Optionally, in other embodiments, the forward urban road network traffic perception optimization model M1 may be obtained by training on the basis of the forward urban road network traffic perception optimization model M obtained in step 2004, that is, the forward urban road network traffic perception optimization model is trained twice; specifically, the forward city road network traffic perception optimization model and the reverse city road network traffic perception optimization model can be trained to increase the diversity of perception results, the trained forward city road network traffic perception optimization model M and the trained reverse city road network traffic perception optimization model are obtained, then the virtual Beidou road traffic data set is obtained based on the trained reverse city road network traffic perception optimization model, and the forward city road network traffic perception optimization model M is subjected to optimization training from the road traffic state determined based on the existing traffic perception strategy to the more accurate road traffic state, so that the forward city road network traffic perception optimization model M1 is obtained.
In the current related art, the training of the urban road network traffic perception optimization model usually uses a road traffic state training database with a single reference for parameter optimization, which limits the use of resources and enables the generated optimization to blindly and unreasonably approximate the optimized road traffic perception result.
According to the embodiment of the application, the reverse urban road network traffic perception optimization model can be trained according to a diversity-driven training target so as to increase the diversity of original perception results of a constructed preset road traffic data training set, so that the forward urban road network traffic perception optimization model is enhanced based on road traffic state information determined based on the existing traffic perception strategy in the rich training set, and the optimization quality of the forward urban road network traffic perception optimization model is finally improved by combining with cross entropy loss of a neural network machine urban road network traffic perception optimization model NMT.
The present application can be used in any neural network machine optimization system (e.g., RNNsearch (cyclic neural network search), Transformer, etc.) implemented according to any deep learning framework (e.g., TensorFlow (a symbolic mathematical system programmed according to a dataflow), PyTorch, etc.). Based on a Back-transfer (Back-optimization), enriching the information of road traffic states determined based on the existing traffic perception strategy, thereby improving the quality of the traffic perception optimization model of the forward urban road network.
As can be seen from the above, the present embodiment may be based on the initial road traffic state to be perceived sent by the server receiving terminal, where the initial road traffic state includes at least one initial road traffic state attribute; acquiring a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first city road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, the virtual Beidou road traffic data sets comprise second road traffic parameters, the second urban road network traffic perception optimization model is trained according to a second practical Beidou road traffic data set to obtain first road traffic parameters, one second practical Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, and the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters; the server outputs at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized to the corresponding road traffic state attribute to be compared based on a first urban road network traffic perception optimization model and according to road traffic attribute linear representation; determining a target comparison road surface traffic state attribute corresponding to each initial road surface traffic state attribute from road surface traffic state attributes to be compared according to a first confidence coefficient based on a first city road network traffic perception optimization model; based on a first urban road network traffic perception optimization model, carrying out standardized output on each target comparison road traffic state attribute to obtain a target road traffic state after the initial road traffic state is optimized; and the server sends the target road traffic state to the terminal. According to the embodiment of the application, the second urban road network traffic perception optimization model can be trained according to a plurality of auxiliary first road traffic parameters, the diversity of the constructed virtual Beidou road traffic data set is increased, richer original information is provided for the training of the first urban road network traffic perception optimization model, and the optimization quality is improved.
Referring to fig. 3, an embodiment of the invention provides a big dipper based urban large area road network traffic sensing system 110, including:
an obtaining module 1101, configured to obtain an initial road traffic state to be perceived, where the initial road traffic state includes at least one initial road traffic state attribute; obtaining a linear representation of road traffic attributes corresponding to at least one initial road traffic state attribute in initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, each virtual Beidou road traffic data set comprises a second road traffic parameter and a first road traffic parameter obtained by optimizing the second road traffic parameter by the second urban road network traffic perception optimization model, the second urban road network traffic perception optimization model replaces weights of preset second urban road network traffic perception optimization models according to a second actual Beidou road traffic data set to construct, one second actual road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters, and the weight of the preset second urban road network traffic perception optimization model is determined according to the optimization results of the preset second urban road network traffic perception optimization model on the target second road traffic parameters and second loss function results among the auxiliary first road traffic parameters corresponding to the target second road traffic parameters.
The output module 1102 is configured to output, based on the first urban road network traffic perception optimization model, at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute according to the road traffic attribute linear representation, and a first confidence degree that each initial road traffic state attribute is optimized to a corresponding road traffic state attribute to be compared.
The optimization module 1103 is configured to determine, based on the first urban road network traffic perception optimization model, a target comparison road traffic state attribute corresponding to each initial road traffic state attribute from road traffic state attributes to be compared according to the first confidence; and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state.
It should be noted that, the implementation principle of the Beidou-based urban large-area road network traffic sensing system 110 may refer to the implementation principle of the Beidou-based urban large-area road network traffic sensing method, and details are not repeated here. It should be understood that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the obtaining module 1101 may be a processing element separately set up, or may be implemented by being integrated into a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and the processing element of the apparatus calls and executes the functions of the obtaining module 1101. The other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.
The embodiment of the present invention provides a computer device 100, where the computer device 100 includes a processor and a nonvolatile memory storing computer instructions, and when the computer instructions are executed by the processor, the computer device 100 executes the foregoing beidou-based urban large area road network traffic sensing system 110. As shown in fig. 4, fig. 4 is a block diagram of a computer device 100 according to an embodiment of the present invention. The computer device 100 comprises a beidou-based urban large area network traffic perception system 110, a memory 111, a processor 112 and a communication unit 113.
The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations are possible in light of the above teaching.

Claims (10)

1. The big dipper based urban large-area road network traffic perception method is characterized by comprising the following steps:
acquiring an initial road traffic state to be perceived, wherein the initial road traffic state comprises at least one initial road traffic state attribute;
obtaining a road traffic attribute linear representation corresponding to at least one initial road traffic state attribute in the initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, each virtual Beidou road traffic data set comprises a second road traffic parameter and a first road traffic parameter obtained by optimizing the second road traffic parameter through a second urban road network traffic perception optimization model, the second urban road traffic perception optimization model replaces the weight of a preset second urban traffic perception optimization model according to a second actual Beidou road traffic data set to construct, and one second actual Beidou road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters, and the weight of a preset second urban road network traffic perception optimization model is determined according to the optimized results of the preset second urban road network traffic perception optimization model on the target second road traffic parameters and second loss function results among the auxiliary first road traffic parameters corresponding to the target second road traffic parameters;
based on the first urban road network traffic perception optimization model, according to the road traffic attribute linear representation, outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute, and a first confidence coefficient of each initial road traffic state attribute which is optimized to be the corresponding road traffic state attribute to be compared;
determining a target contrast road surface traffic state attribute corresponding to each initial road surface traffic state attribute from the road surface traffic state attributes to be contrasted according to the first confidence coefficient based on the first urban road network traffic perception optimization model;
and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attributes of the comparison road traffic states of the targets to obtain the optimized target road traffic state of the initial road traffic state.
2. The method of claim 1, wherein the outputting at least two road traffic state attributes to be compared corresponding to each initial road traffic state attribute according to the road traffic attribute linear representation based on the first city road network traffic perception optimization model, and the optimizing each initial road traffic state attribute to a first confidence degree of the corresponding road traffic state attribute to be compared comprises:
on the basis of the correlation corresponding relation between a first road traffic attribute linear space of an initial perception strategy and a second road traffic attribute linear space of an optimized perception strategy in the first urban road network traffic perception optimization model, performing benchmarking on the road traffic attribute linear representation to the second road traffic attribute linear space to obtain a reference road traffic attribute linear representation;
according to the road traffic attribute linear representation of the undetermined road traffic state attribute value of the optimized perception strategy in the second road traffic attribute linear space and the space distance linearly represented by the reference road traffic attribute, determining at least two undetermined contrast road traffic state attributes corresponding to each initial road traffic state attribute in the undetermined road traffic state attribute value, and optimizing each initial road traffic state attribute into a first confidence coefficient of the corresponding undetermined contrast road traffic state attribute.
3. The method of claim 1, wherein prior to obtaining the initial road traffic condition to be perceived, further comprising:
acquiring second road traffic parameters and a plurality of first actual Beidou road traffic data sets, wherein each first actual Beidou road traffic data set comprises configured first road traffic parameters and configured second road traffic parameters which are matched with each other;
optimizing the second road traffic parameters based on a second urban road network traffic perception optimization model to obtain optimized first road traffic parameters, and forming a virtual Beidou road traffic data set by the optimized first road traffic parameters and the corresponding second road traffic parameters;
determining the virtual Beidou road traffic data set and the first actual Beidou road traffic data set as actual Beidou road traffic data set training samples of the first urban road network traffic perception optimization model, wherein the configured first road traffic parameters and the optimized first road traffic parameters are first road traffic parameter samples, and the configured second road traffic parameters and the optimized second road traffic parameters are second road traffic parameter samples;
optimizing the first road traffic parameter sample based on a first urban road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimization perception strategy on the first urban road network traffic perception optimization model according to a first loss function result between the predicted second road traffic parameter and the second road traffic parameter sample corresponding to the first road traffic parameter sample.
4. The method according to claim 3, wherein before optimizing the second road traffic parameter based on the second city road network traffic perception optimization model to obtain the optimized first road traffic parameter, the method further comprises:
acquiring a second actual Beidou road traffic data set;
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in the target second road traffic parameters based on a preset second urban road network traffic perception optimization model;
according to the road traffic attribute linear representation corresponding to the target road traffic state attribute, outputting at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute and a second confidence coefficient of each target road traffic state attribute optimized to the corresponding candidate initial road traffic attribute value;
according to the second confidence coefficient, determining a target initial road traffic attribute value corresponding to each target road traffic state attribute from the candidate initial road traffic attribute values;
carrying out standardized output on each target initial road traffic attribute value to obtain a predicted first road traffic parameter after the target second road traffic parameter is optimized;
calculating a second loss function result between the predicted first road traffic parameter of the same target second road traffic parameter and each corresponding auxiliary first road traffic parameter;
and replacing the weight of the preset second urban road network traffic perception optimization model according to the second loss function result to obtain a second urban road network traffic perception optimization model.
5. The method of claim 4, wherein outputting at least two candidate initial road traffic property values corresponding to each of the target road traffic state properties according to the linear representation of road traffic property corresponding to the target road traffic state property, and wherein optimizing each of the target road traffic state properties to a second confidence level of the corresponding candidate initial road traffic property values comprises:
acquiring an initial road traffic attribute value set, wherein the initial road traffic attribute value set comprises a plurality of initial road traffic attribute values;
performing benchmarking on the road traffic attribute linear representation of the target road traffic state attribute into a first road traffic attribute linear space based on the incidence corresponding relation between a second road traffic attribute linear space of an optimized perception strategy and a first road traffic attribute linear space of an initial perception strategy in the preset second urban road network traffic perception optimization model to obtain a first reference road traffic attribute linear representation;
according to the linear representation of the road traffic attributes of the initial road traffic attribute values in the linear space of the first road traffic attribute and the space distance of the linear representation of the first reference road traffic attribute, determining at least two candidate initial road traffic attribute values corresponding to each target road traffic state attribute in the initial road traffic attribute values, and optimizing each target road traffic state attribute into a second confidence coefficient of the corresponding candidate initial road traffic attribute value.
6. The method according to claim 5, wherein said calculating a second loss function result between the predicted first road traffic parameter and each corresponding auxiliary first road traffic parameter for the same target second road traffic parameter comprises:
for the same target second road traffic parameter, calculating a candidate initial road traffic attribute value corresponding to the target road traffic state attribute of the same target second road traffic parameter, and calculating a first matching degree between the target road traffic attribute values corresponding to the target road traffic state attribute in the auxiliary first road traffic parameter corresponding to the target second road traffic parameter;
calculating road traffic attribute loss between candidate initial road traffic attribute values corresponding to each target road traffic state attribute in the target second road traffic parameters and target road traffic attribute values according to the first matching degree and the second confidence degree;
and carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a second loss function result between the predicted first road traffic parameter of the target second road traffic parameter and each corresponding auxiliary first road traffic parameter.
7. The method according to claim 3, wherein before optimizing said first road traffic parameter sample based on said first city road network traffic perception optimization model to obtain an optimized predicted second road traffic parameter, further comprising:
acquiring a third practical Beidou road traffic data set, wherein one third practical Beidou road traffic data set comprises a target first road traffic parameter and a plurality of corresponding auxiliary second road traffic parameters, and the auxiliary second road traffic parameters are optimized road traffic perception results of the target first road traffic parameter;
acquiring road traffic attribute linear representation corresponding to each target road traffic state attribute in the target first road traffic parameters based on a preset first urban road network traffic perception optimization model;
obtaining an optimized road traffic attribute value set, wherein the optimized road traffic attribute value set comprises a plurality of optimized road traffic attribute values;
performing benchmarking on the road traffic attribute linear representation of the target road traffic state attribute to a second road traffic attribute linear space based on the incidence corresponding relation between the first road traffic attribute linear space of the initial perception strategy and the second road traffic attribute linear space of the optimized perception strategy in the preset first city road network traffic perception optimization model to obtain a second reference road traffic attribute linear representation;
determining at least two candidate optimized road traffic attribute values corresponding to each target road traffic state attribute in the optimized road traffic attribute values and a third confidence degree that each target road traffic state attribute is optimized to a corresponding candidate optimized road traffic attribute value according to the linear road traffic attribute representation of the optimized road traffic attribute values in the second road traffic attribute linear space and the space distance of the linear road traffic attribute representation of the second reference road traffic attribute;
according to the third confidence coefficient, determining a target optimized road traffic attribute value corresponding to each target road traffic state attribute from the candidate optimized road traffic attribute values;
carrying out standardized output on each target optimized road traffic attribute value to obtain a predicted target second road traffic parameter after the target first road traffic parameter is optimized;
calculating a candidate optimized road traffic attribute value corresponding to a target road traffic state attribute of the same target first road traffic parameter, and calculating a second matching degree between the target road traffic attribute values corresponding to the target road traffic state attribute in an auxiliary second road traffic parameter corresponding to the target first road traffic parameter;
calculating road traffic attribute loss between the candidate optimized road traffic attribute value corresponding to each target road traffic state attribute in the target first road traffic parameter and the target road traffic attribute value according to the second matching degree and the third confidence degree;
carrying out standardized output on the road traffic attribute loss corresponding to each target road traffic state attribute to obtain a third loss function result between a predicted target second road traffic parameter of the target first road traffic parameter and each corresponding auxiliary second road traffic parameter;
and replacing the weight of the preset first city road network traffic perception optimization model according to the third loss function result to obtain a first city road network traffic perception optimization model.
8. The method according to claim 3, wherein said optimizing said first road traffic parameter sample based on said first city road traffic perception optimization model to obtain an optimized predicted second road traffic parameter, and performing optimization training from an initial perception strategy to an optimized perception strategy on said first city road traffic perception optimization model according to a first loss function result between said predicted second road traffic parameter and said second road traffic parameter sample corresponding to said first road traffic parameter sample, comprises:
acquiring road traffic attribute linear representation corresponding to each road traffic state attribute in the first road traffic parameter sample based on a first urban road network traffic perception optimization model;
according to the road traffic attribute linear representation, outputting at least two road traffic state attributes to be selected corresponding to each road traffic state attribute, and optimizing each road traffic state attribute into a fourth confidence coefficient of the corresponding road traffic state attribute to be selected;
according to the fourth confidence coefficient, determining a target road traffic state attribute corresponding to each road traffic state attribute from the road traffic state attributes to be selected, and performing standardized output on each target road traffic state attribute to obtain a predicted second road traffic parameter;
calculating a first loss function result between the predicted second road traffic parameter and a second road traffic parameter sample corresponding to the first road traffic parameter sample;
and replacing the weight of the first urban road network traffic perception optimization model according to the first loss function result to obtain the trained first urban road network traffic perception optimization model.
9. The utility model provides an urban large tracts of land road network traffic perception system based on big dipper which characterized in that includes:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring an initial road traffic state to be perceived, and the initial road traffic state comprises at least one initial road traffic state attribute; acquiring a road traffic attribute linear representation corresponding to the at least one initial road traffic state attribute in the initial road traffic states based on a first urban road network traffic perception optimization model, wherein the first urban road network traffic perception optimization model is obtained by training according to a plurality of virtual Beidou road traffic data sets, each virtual Beidou road traffic data set comprises a second road traffic parameter and a first road traffic parameter obtained by optimizing the second road traffic parameter through a second urban road network traffic perception optimization model, the second urban road network traffic perception optimization model replaces the preset weight of the second urban road network traffic perception optimization model according to a second actual Beidou road traffic data set to construct, and one second actual road traffic data set comprises a target second road traffic parameter and a plurality of auxiliary first road traffic parameters, the auxiliary first road traffic parameters are optimized road traffic perception results of the target second road traffic parameters, and the weight of a preset second urban road network traffic perception optimization model is determined according to the optimized results of the preset second urban road network traffic perception optimization model on the target second road traffic parameters and second loss function results among the auxiliary first road traffic parameters corresponding to the target second road traffic parameters;
the output module is used for outputting at least two road traffic state attributes to be determined and compared corresponding to each initial road traffic state attribute and a first confidence coefficient of each initial road traffic state attribute which is optimized into the corresponding road traffic state attribute to be determined and compared based on the first urban road network traffic perception optimization model and according to the road traffic attribute linear representation;
the optimization module is used for determining a target contrast road traffic state attribute corresponding to each initial road traffic state attribute from the road traffic state attributes to be contrasted based on the first city road network traffic perception optimization model and according to the first confidence coefficient; and based on the first urban road network traffic perception optimization model, carrying out standardized output on the attribute of each target contrast road traffic state to obtain the target road traffic state after the initial road traffic state is optimized.
10. The Beidou-based urban large-area road network traffic perception cloud platform is characterized by comprising computer equipment and a user terminal, wherein the computer equipment is used for executing the Beidou-based urban large-area road network traffic perception method in any one of claims 1-8, and the user terminal is used for receiving a target road traffic state.
CN202210662467.7A 2022-06-13 2022-06-13 Beidou-based urban large-area road network traffic sensing method and system and cloud platform Active CN114743379B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210662467.7A CN114743379B (en) 2022-06-13 2022-06-13 Beidou-based urban large-area road network traffic sensing method and system and cloud platform

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210662467.7A CN114743379B (en) 2022-06-13 2022-06-13 Beidou-based urban large-area road network traffic sensing method and system and cloud platform

Publications (2)

Publication Number Publication Date
CN114743379A true CN114743379A (en) 2022-07-12
CN114743379B CN114743379B (en) 2022-09-06

Family

ID=82286886

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210662467.7A Active CN114743379B (en) 2022-06-13 2022-06-13 Beidou-based urban large-area road network traffic sensing method and system and cloud platform

Country Status (1)

Country Link
CN (1) CN114743379B (en)

Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH07220193A (en) * 1994-01-28 1995-08-18 Nagoya Denki Kogyo Kk Traffic condition discriminating device
CN102568230A (en) * 2010-12-20 2012-07-11 财团法人工业技术研究院 Real-time traffic network sensing system and method
CN102592447A (en) * 2011-12-20 2012-07-18 浙江工业大学 Method for judging road traffic state of regional road network based on fuzzy c means (FCM)
CN102750824A (en) * 2012-06-19 2012-10-24 银江股份有限公司 Urban road traffic condition detection method based on voting of network sorter
CN104574967A (en) * 2015-01-14 2015-04-29 合肥革绿信息科技有限公司 City large-area road network traffic sensing method based on plough satellite
CN104834977A (en) * 2015-05-15 2015-08-12 浙江银江研究院有限公司 Traffic alarm condition level predication method based on distance metric learning
CN104933856A (en) * 2014-03-17 2015-09-23 径卫视觉科技（上海）有限公司 Real-time evaluation system and method for road conditions
CN106157621A (en) * 2016-07-26 2016-11-23 中国科学院自动化研究所 A kind of Intelligent Road based on data analysis management system
CN106251625A (en) * 2016-08-18 2016-12-21 上海交通大学 Three-dimensional urban road network global state Forecasting Methodology under big data environment
CN110060489A (en) * 2019-03-27 2019-07-26 浙江工业大学 A kind of traffic signal timing scheme recommended method neural network based
WO2020040411A1 (en) * 2018-08-21 2020-02-27 한국과학기술정보연구원 Device for predicting traffic state information, method for predicting traffic state information, and storage medium for storing program for predicting traffic state information
CN111680745A (en) * 2020-06-08 2020-09-18 青岛大学 Burst congestion judging method and system based on multi-source traffic big data fusion
WO2020220439A1 (en) * 2019-04-29 2020-11-05 东北大学 Highway traffic flow state recognition method based on deep neural network
CN111915878A (en) * 2019-05-09 2020-11-10 奥迪股份公司 Method and device for predicting road traffic state, computer device and storage medium
CN112465253A (en) * 2020-12-09 2021-03-09 重庆邮电大学 Method and device for predicting links in urban road network
CN113160556A (en) * 2021-03-12 2021-07-23 北京邮电大学 Urban road network dynamic division method and device, computer equipment and storage medium
CN113887704A (en) * 2021-09-29 2022-01-04 南方科技大学 Traffic information prediction method, device, equipment and storage medium
CN114419878A (en) * 2021-12-22 2022-04-29 银江技术股份有限公司 Method, electronic device and storage medium for urban road network global traffic state prediction

Patent Citations (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH07220193A (en) * 1994-01-28 1995-08-18 Nagoya Denki Kogyo Kk Traffic condition discriminating device
CN102568230A (en) * 2010-12-20 2012-07-11 财团法人工业技术研究院 Real-time traffic network sensing system and method
CN102592447A (en) * 2011-12-20 2012-07-18 浙江工业大学 Method for judging road traffic state of regional road network based on fuzzy c means (FCM)
CN102750824A (en) * 2012-06-19 2012-10-24 银江股份有限公司 Urban road traffic condition detection method based on voting of network sorter
CN104933856A (en) * 2014-03-17 2015-09-23 径卫视觉科技（上海）有限公司 Real-time evaluation system and method for road conditions
CN104574967A (en) * 2015-01-14 2015-04-29 合肥革绿信息科技有限公司 City large-area road network traffic sensing method based on plough satellite
CN104834977A (en) * 2015-05-15 2015-08-12 浙江银江研究院有限公司 Traffic alarm condition level predication method based on distance metric learning
CN106157621A (en) * 2016-07-26 2016-11-23 中国科学院自动化研究所 A kind of Intelligent Road based on data analysis management system
CN106251625A (en) * 2016-08-18 2016-12-21 上海交通大学 Three-dimensional urban road network global state Forecasting Methodology under big data environment
WO2020040411A1 (en) * 2018-08-21 2020-02-27 한국과학기술정보연구원 Device for predicting traffic state information, method for predicting traffic state information, and storage medium for storing program for predicting traffic state information
CN110060489A (en) * 2019-03-27 2019-07-26 浙江工业大学 A kind of traffic signal timing scheme recommended method neural network based
WO2020220439A1 (en) * 2019-04-29 2020-11-05 东北大学 Highway traffic flow state recognition method based on deep neural network
CN111915878A (en) * 2019-05-09 2020-11-10 奥迪股份公司 Method and device for predicting road traffic state, computer device and storage medium
CN111680745A (en) * 2020-06-08 2020-09-18 青岛大学 Burst congestion judging method and system based on multi-source traffic big data fusion
CN112465253A (en) * 2020-12-09 2021-03-09 重庆邮电大学 Method and device for predicting links in urban road network
CN113160556A (en) * 2021-03-12 2021-07-23 北京邮电大学 Urban road network dynamic division method and device, computer equipment and storage medium
CN113887704A (en) * 2021-09-29 2022-01-04 南方科技大学 Traffic information prediction method, device, equipment and storage medium
CN114419878A (en) * 2021-12-22 2022-04-29 银江技术股份有限公司 Method, electronic device and storage medium for urban road network global traffic state prediction

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
汪洋: "无线传感器网络与群智感知网络的优化与融合", 《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》 *

Also Published As

Publication number Publication date
CN114743379B (en) 2022-09-06

Similar Documents

Publication Publication Date Title
CN106471851B (en) Learning model based device positioning
CN111753076B (en) Dialogue method, dialogue device, electronic equipment and readable storage medium
CN111209215A (en) Application program testing method and device, computer equipment and storage medium
CN112556692B (en) Vision and inertia odometer method and system based on attention mechanism
CN115616937B (en) Automatic driving simulation test method, device, equipment and computer readable medium
CN112699765A (en) Method and device for evaluating visual positioning algorithm, electronic equipment and storage medium
CN112085056A (en) Target detection model generation method, device, equipment and storage medium
CN113379045A (en) Data enhancement method and device
CN114743379B (en) Beidou-based urban large-area road network traffic sensing method and system and cloud platform
CN115577797B (en) Federal learning optimization method and system based on local noise perception
CN116781343A (en) Terminal credibility assessment method, device, system, equipment and medium
CN115016950B (en) Data analysis method and system based on multithreading model
CN109581280A (en) The adaptive tuning on-line method, system and device of terminal
CN114543797A (en) Pose prediction method and apparatus, device, and medium
CN115858911A (en) Information recommendation method and device, electronic equipment and computer-readable storage medium
CN109743203B (en) Distributed service security combination system and method based on quantitative information flow
CN104350491A (en) Data sampling method and data sampling device
CN105913365A (en) Environment monitoring GIS information transmission unit
CN115049730B (en) Component mounting method, component mounting device, electronic apparatus, and storage medium
CN115577318B (en) Semi-physical-based data fusion evaluation method, system, equipment and storage medium
CN109726847A (en) Position predicting method, device, electronic equipment and readable storage medium storing program for executing
JP7372377B2 (en) Road information determination method and device, electronic equipment, storage medium, and computer program
CN114840764B (en) Big data mining method serving user interest analysis and cloud AI deployment system
CN114926655B (en) Training method and position determining method of geographic and visual cross-mode pre-training model
CN114998684B (en) Training method and positioning adjustment method for geographic and visual cross-mode pre-training model

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
CP02 Change in the address of a patent holder
CP02 Change in the address of a patent holder

Address after: One of the 11th Floor, Building 4, Shunde Zhichuang Park, No. 68, Defu Road, Xingtan Town, Shunde District, Foshan City, Guangdong Province, 528000 (Residence declaration)

Patentee after: Guangdong Bangsheng Beidou Technology Co.,Ltd.

Address before: 528400 one of the eighth floor, phase I, Fuye Plaza, No. 16, Fuhua Road, West District, Zhongshan City, Guangdong Province

Patentee before: Guangdong Bangsheng Beidou Technology Co.,Ltd.

PE01 Entry into force of the registration of the contract for pledge of patent right
PE01 Entry into force of the registration of the contract for pledge of patent right

Denomination of invention: A Traffic Perception Method, System, and Cloud Platform for Urban Large Area Road Network Based on Beidou

Effective date of registration: 20231113

Granted publication date: 20220906

Pledgee: Shunde Guangdong rural commercial bank Limited by Share Ltd. Daliang branch

Pledgor: Guangdong Bangsheng Beidou Technology Co.,Ltd.

Registration number: Y2023980065299

EE01 Entry into force of recordation of patent licensing contract
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20220712

Assignee: Guangdong Beidou Car Service Technology Co.,Ltd.

Assignor: Guangdong Bangsheng Beidou Technology Co.,Ltd.

Contract record no.: X2024980000443

Denomination of invention: A Traffic Perception Method, System, and Cloud Platform for Urban Large Area Road Network Based on Beidou

Granted publication date: 20220906

License type: Common License

Record date: 20240110