CN109410577B - Self-adaptive traffic control subarea division method based on space data mining - Google Patents
Self-adaptive traffic control subarea division method based on space data mining Download PDFInfo
- Publication number
- CN109410577B CN109410577B CN201811332450.5A CN201811332450A CN109410577B CN 109410577 B CN109410577 B CN 109410577B CN 201811332450 A CN201811332450 A CN 201811332450A CN 109410577 B CN109410577 B CN 109410577B
- Authority
- CN
- China
- Prior art keywords
- intersection
- traffic
- flow
- influence
- weight
- 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.)
- Active
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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
- G08G1/0145—Measuring and analyzing of parameters relative to traffic conditions for specific applications for active traffic flow control
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
Abstract
A self-adaptive traffic control subarea division method based on spatial data mining belongs to the technical field of traffic. The invention aims to solve the problem that the travel efficiency of a control area divided by the existing control subarea dividing method is low. Firstly, extracting characteristics of high and low peak time periods and high, medium and low flow thresholds, and selecting traffic information characteristics influencing division of a control subarea; and then establishing an influence weight calculation model based on space-time characteristics to obtain the weight of the characteristic r, and finally realizing the division of the traffic control subareas based on a weighted community discovery algorithm of dynamic modularity division. The method is suitable for dividing the traffic control subareas.
Description
Technical Field
The invention belongs to the technical field of traffic, and particularly relates to a traffic control subregion division method.
Background
With the increasing development of traffic, traffic congestion becomes a common phenomenon in daily travel, and the living standard and quality of people are seriously affected, so that the traffic congestion relief becomes an important difficult problem to be solved urgently in intelligent traffic development. In the existing control theory and algorithm, signal control is a common technical means for relieving congestion, has the most obvious improvement effect and the highest cost performance, and is an important means for improving the traffic control level and managing the travel efficiency at present.
The control subarea division is a process of dividing the regional road network into different control subareas according to the dynamic traffic characteristics and static geographic characteristics of each signal intersection by using methods such as a relevance model, fuzzy control, machine learning, optimization theory and the like. The method for extracting the traffic information features mostly adopts a spatial data mining method.
And dividing a traffic control subarea based on fuzzy control. The fuzzy inference method of the coordination coefficient is determined by comprehensively considering the influence of 5 factors such as traffic flow, intersection distance, period, traffic flow composition, traffic flow discreteness and the like, and a judgment basis for dividing the control subareas by utilizing the coordination coefficient is provided. On the basis of the relevance calculation method of fuzzy logic, a traffic control subregion partition model taking key intersections and the relevance of the intersections as a judgment basis is constructed, and a genetic algorithm is utilized to improve the model.
And dividing a traffic control subarea based on machine learning. The dynamic control subarea dividing method based on the combination of static division and dynamic division improves the principle of maximum-minimum intersection distance, and improves the real-time performance of the control subarea division by integrating the short-time traffic flow prediction result. A traffic control subarea division model based on flow, road section length and traffic period is provided by fusing a fuzzy control technology and a neural network model, a fuzzy inference logic is determined through the neural network to calculate a coordination coefficient of an adjacent intersection, and the control subareas are divided according to the size of the coordination coefficient.
A traffic control subarea dividing method based on a relevance model. The method for quantifying the relevance of the comprehensive characteristics of the road based on the fleet discrete model establishes an index correction formula to provide decision support for whether traffic control sub-areas are combined or not by considering factors such as traffic flow, intersection distance, fleet discrete characteristics and the like. The road network division method based on the intersection association degree and the road section association degree realizes the induction and control cooperation oriented road network association degree modeling by analyzing the influence factors of the road network region division of traffic control and traffic induction and introducing the traffic state discrimination indexes while considering the traffic flow correlation, thereby realizing the division of the cooperative traffic subareas. A calculation formula of the combined relevance of the multiple intersections is given on the basis of the calculation model of the relevance of the adjacent intersections. A coordinated control subarea division model is established by defining a solution set space, constraint conditions and evaluation criteria of a control subarea division scheme.
The extraction of the traffic information features is mainly processed aiming at the track data. A spatial trajectory is a trajectory generated by a moving object in geospatial space, typically represented by a series of chronologically arranged points, where each point is composed of a set of geospatial coordinates and a time stamp. The practical problems that are solved by applying different spatial mining techniques to trajectory data to obtain information are becoming more and more widespread. A traffic jam prediction model based on a spatial data mining technology and a semantic network technology improves the accuracy and consistency of traffic jam prediction by extracting comprehensive information such as weather, road engineering, urban events and the like. And acquiring road background characteristics from continuous multi-frame images by a background image analysis method to extract a blocking event, and analyzing corner characteristics of a road by a space-time data mining method to detect traffic blocking.
The existing research has achieved a certain achievement on the research of extracting traffic information characteristics and dividing the control subarea, but due to the influence of factors such as complexity and variability of an urban road network, redundancy of the mineable traffic information data volume and the like, the dynamic dividing technology of the traffic control subarea is difficult to have universal applicability. And the topological structure characteristic of the traffic network is ignored by the selected traffic information characteristic when the adjacent intersection coordination coefficient is calculated.
Disclosure of Invention
The invention aims to solve the problem that the travel efficiency of a control area divided by the existing control subarea dividing method is low.
The self-adaptive traffic control subarea division method based on the space data mining comprises the following steps:
step 1, extracting characteristics of high and low peak time periods and high, medium and low flow thresholds;
step 2, selecting traffic information characteristics influencing division of the control subarea;
the traffic information features include: traffic flow influence characteristic TF, intersection distance influence characteristic IF, traffic flow composition influence characteristic CI and period influence characteristic CE;
step 3, establishing an influence weight calculation model based on space-time characteristics to obtain a weight L 'of the characteristics r'r，r∈[1,4]The characteristic labels in the traffic information characteristics represent traffic flow influence characteristics TF, intersection distance influence characteristics IF, traffic flow composition influence characteristics CI and cycle influence characteristics CE;
step 4, realizing the division of the traffic control subareas by a weighted community discovery algorithm based on dynamic modularity division, wherein the specific process comprises the following steps:
step 4.1, obtaining a calculation coordination coefficient CC between every two intersections by performing product accumulation on each feature weight calculated by the influence weight calculation model based on the space-time featuresijAs the weight of the road section, wherein R is the values of IF, CI, CE and TF respectively;
step 4.2, when initializing a community matrix, each node in the graph is an independent community, the number of communities in the initial state is consistent with the number of nodes, the adjacent matrix represents the weighted undirected graph, and the degree weight sum of each node is calculated;
4.3, each node and adjacent nodes try to carry out combined community operation to obtain a modularity variation value delta Q before and after distribution, record a community combination state with the maximum delta Q value and the delta Q value larger than 0, and distribute the nodes to communities where the adjacent nodes are located, otherwise, the nodes are kept unchanged;
4.4, completing one merging operation until the modularity change value caused by the community structure change of the weighted undirected graph is smaller than a threshold value, namely the modularity value is maximum;
step 4.5, compressing the graph according to the recorded combined community state, reconstructing all nodes in the same community into a new node, creating a new community matrix, converting the weight of edges among the nodes in the community into the weight of a new node ring, and converting the weight of edges among the communities into the weight of edges among the new nodes;
and 4.6, repeating the calculation process from the step 4.2 to the step 4.5 until the modularity variation value of the whole graph is smaller than the threshold value, and finally realizing the division of the traffic control subareas.
Further, the step 1 of performing feature extraction of the high and low peak periods and the high, medium and low flow thresholds includes the following steps:
step 1.1, calculating traffic flows in different traveling directions passing through the intersection, namely calculating the traffic flows of roads which enter the intersection from one of the two intersected roads and then travel to different directions;
step 1.2, dividing high and low peak time periods according to the traffic flow corresponding to each time interval:
after the traffic flows in different advancing directions of the intersection are calculated, a mathematical model for solving high, medium and low flow thresholds is constructed on the basis of high and low peak time interval division; the design sets dynamic weight parameters s for different high and low peak time intervals according to real-time traffic flowjDefining weight formula sj
sj＝α/γ
Wherein alpha is the traffic flow of the intersection at the current peak time, and gamma is the sum of the traffic flows of all the peak time of the intersection;
then establishing the traffic flow of the single intersection in the peak period:
wherein j belongs to the peak time period, i belongs to the [1,12] which is the traveling direction of the vehicle, and num represents the number of the peak time period;
establishing: tuple representation of high and low peak time interval flow of single intersection, and high peak time interval flow tuple D of single intersectionthIs composed of
Dth＝[dth1,dth2,...,dth12]
Similar to the above method, the peak-low time interval traffic tuple D of a single intersection can be obtainedtlIs composed of
Dtl＝[dtl1,dtl2,...,dtl12]
Step 1.3, ordering the traffic flow in the same direction of a single intersection, wherein the flow matrix of each intersection is
Wherein q isIJThe traffic flow value after the traffic flow sorting is shown, wherein I is 12, 1 to I represent the traveling direction, J represents a time interval, and the time interval is marked as a time interval;
step 1.4 according to DthAnd DtlD in the same direction of travelthiAnd dtliDividing time intervals in the same traveling direction in the sequenced flow matrix Q into low-flow areas S corresponding to the traffic flowLMedium flow rate region SMHigh flow area SH：
The lengths of the three sequences are respectively marked as l1, l2 and l 3; for the i traveling direction, respectively take SL，SMIs used as the middle and low flow representative value r in the i traveling directionliAnd the representative value r of the sum flowmiTaking SH70 decimals of (a) as a high flow rate representative value r in the ith directionhi：
Representing the flow threshold matrix as R (I E [1,12 ]);
further, the specific process of step 1.1 includes the following steps:
step 1.1.1, marking two crossed roads of the intersection as a first road and a second road respectively, and taking two side edges of the first road and the second road as boundaries; the intersection divides two intersecting roads into 5 areas which are respectively an intersection, a front section of a first road intersection, a rear section of the first road intersection, a front section of a second road intersection and a rear section of the second road intersection;
step 1.1.2, judging the original driving direction of the vehicle, namely the direction from which the vehicle enters the intersection;
then, judging the traveling direction of the vehicle according to the subsequent track direction of the current vehicle;
step 1.1.3, judging a time interval to which the current vehicle belongs according to the time tag of the current track point;
step 1.1.4, accumulating the current vehicle to the traffic flow matrix (mu)itλ), i represents a traveling direction, i ∈ [1,12]](ii) a t denotes the number of the time interval, μitA traffic flow representing i a travel direction t time interval; and lambda is the intersection mark.
Further, the traffic flow influence characteristics in step 2 are as follows:
traffic flow influence characteristic TF:
wherein q (t) is the flow of t time interval, sigma is the flow of one direction of t time interval of the intersection, and lambda is the weight factor of different traffic flow directions; the introduction of λ effectively adjusts the calculated TF value to between 0 and 100.
Further, the intersection distance influence characteristics in step 2 are as follows:
setting the distance between adjacent intersections as L;
the maximum-minimum spacing principle is as follows:
maximum spacing principle:
ρ＝avg(C1(t),C2(t))
in the maximum distance principle, RminAverage value of high flow representative values in two adjacent crossroads, NinThe direction of a vehicle which can run in the current green light period; c1(t)、C2(t) respectively representing the signal periods of a front intersection and a back intersection in two adjacent intersections; t issampleIs the length of a time unit, ρ is the average value in the signal period of an adjacent intersection, LmaxRepresents the maximum distance between adjacent intersections, hcarIs the average length of the vehicle, NlaneIs the number of lanes, β vehicle discrete distance, F traffic flow discrete coefficient;
minimum spacing principle:
ρ＝max(C1(t),C2(t))
in the principle of minimum spacing, RmaxLarger value of high flow rate representative value, T, in two adjacent crossroadssampleIs the length of a time unit, ρ is the larger value in the signal period of the adjacent intersection, μ is the anti-congestion factor, LminIndicates the minimum distance between adjacent intersections, hcarIs the vehicle average length;
on the basis of the maximum-minimum principle of the intersection, establishing intersection distance influence characteristics IF:
wherein χ is a balance factor; and l is the length of the current road section.
Further, the traffic flow composition influence characteristics in step 2 are as follows:
the traffic flow composition is an important parameter of traffic information characteristics, the discrete size of traffic flow and the coordination size of upstream and downstream intersections are determined for the proportion of upstream intersection vehicles in the downstream intersection straight vehicles, and the traffic flow composition influence characteristic CI is established:
CI＝O*F
wherein, O is the proportion of the vehicles which go straight at the upstream and continue to go straight at the downstream, the value range [0,1], and F is a traffic flow discrete coefficient.
Further, the cycle impact characteristics of step 2 are as follows:
establishing a periodic impact characteristic CE:
wherein, T1The long green time, T, of the green time in two adjacent crossroads2The green time is the short one of the green time in two adjacent crossroads.
Further, step 3, establishing an influence weight calculation model based on space-time characteristics to obtain a weight L 'of the characteristic r'rThe process of (2) is as follows:
step 3.1, all samples of the characteristic r in the traffic information characteristic are taken to form a set Xr：
Xr＝[x1,x2,...,xn]T
Where n is the total number of samples, r ∈ [1,4 ]]The characteristic labels in the traffic information characteristics represent traffic flow influence characteristics, intersection distance influence characteristics, traffic flow composition influence characteristics and cycle influence characteristics; x is the number of1,x2,...,xnSample representation forms of characteristic values corresponding to the characteristic r respectively;
if a sample point xi'Is another samplePoint xj'Is one of the nearest k sample points, then xi'And xj'An edge exists between the two, and the rule is utilized to respectively establish a feature undirected graph of each feature;
step 3.2, calculating the weight W of each edge in the characteristic undirected graphi'j'Form a weight matrix WrForming an adjacent matrix;
step 3.3, calculating the variance of all samples of the characteristics; drIs a diagonal matrix whose diagonal element value is the sum of the out-of-measure weights of each sample point and is based on DrThe element value in (1) is an intermediate variable X'r：
Dr＝diag(Wr*1) 1＝(1,1,...,1)T
Thereby calculating the variance of all samples of the feature
Step 3.4, knowing WrThe weight matrix is used for carrying out influence weight calculation on the space-time characteristics on the characteristics, and then the calculation can obtain
Lr＝Dr-Wr
xri'、xrj'Respectively representing sample points x corresponding to the features ri'、xj'；
The weight of the feature r is therefore:
further, k is 5.
The invention has the following effects:
the traffic control subarea division scheme optimizes the running time of 5 trips to different degrees, and the longer the distance of the trips is, the more obvious the optimization effect is. The optimization results of multiple strokes are integrated, the average stroke is shortened by 22.4% of travel time, and the travel efficiency is improved. The traffic control subarea division mathematical model provided by the invention has a good effect on optimizing the vehicle travel, effectively reduces the travel time of the vehicle in long-distance travel, shortens the delay time of the vehicle in travel, increases the saturation and the utilization rate of roads, effectively reduces the parking times, improves the travel speed to a certain extent and comprehensively optimizes the travel efficiency and road conditions under the same travel distance.
Meanwhile, the control subarea division method can be suitable for dividing the control subareas of the areas corresponding to various road conditions, and the application rate is higher.
Drawings
FIG. 1 is a schematic view illustrating the division and determination of traffic flow direction zones;
FIG. 2 is a one week traffic flow profile;
FIG. 3 is a one day traffic flow profile;
FIG. 4 is a vehicle trajectory diagram;
FIG. 5 is a real map;
FIG. 6 is a cross-road port area view;
FIG. 7 is a traffic characteristic weight coefficient plot;
FIG. 8 is a diagram of control sub-division results;
FIG. 9 is a diagram of the dynamic modularity partitioned Louvain calculation process;
FIG. 10 is a comparison graph of road simulation and actual intersection location;
fig. 11 is a signal cycle configuration diagram.
Detailed Description
The first embodiment is as follows:
the self-adaptive traffic control subarea division method based on the space data mining comprises the following steps:
step 1, extracting characteristics of high and low peak time periods and high, medium and low flow thresholds:
the high, medium and low flow threshold values are important division boundaries for determining traffic states of the intersection in different directions in the current time period, and are important parameters for realizing the flow principle in a later-stage traffic control sub-area model. The precondition for calculating the high, medium and low flows is that the data mining method is needed to realize the flow statistics of different directions of the intersection at different time intervals.
Step 1.1, calculating the traffic flow of different traveling directions passing through the intersection, namely calculating the traffic flow of the road which enters the intersection from one of the two intersected roads and then drives to the different directions:
step 1.1.1, respectively marking two crossed roads of the intersection as a first road and a second road, wherein two side edges of the first road and the second road are used as boundaries, and the boundary labels of the roads are L1, L2, L3 and L4; the intersection divides two intersecting roads into 5 regions, which are respectively an intersection, a front section of a first road intersection, a rear section of the first road intersection, a front section of a second road intersection and a rear section of the second road intersection, and are marked as a region of number one-fifthly, as shown in figure 1;
step 1.1.2, judging the original driving direction of the vehicle, namely the direction from which the vehicle enters the intersection shown by the fifth area;
then, judging the traveling direction of the vehicle according to the subsequent track direction of the current vehicle; the vehicle has left, forward and right subsequent track directions after the front section of each road enters the intersection, so each road has three subsequent track directions, and the four direction roads connected with the intersection have 12 advancing directions in total, as shown by the advancing directions 1-12 (arrows) in fig. 1;
the judgment process can be judged according to the position corresponding to the location coordinate Loc (xi ', yi') corresponding to the front section of the intersection of each road and the position corresponding to the location coordinate corresponding to the subsequent track, which is described with reference to fig. 1,
wherein, Loc (x)j,xj) Is Loc (x)i,xi) The successor nodes of (1).
Similarly, the traffic flow data in other directions can be judged;
step 1.1.3, judging a time interval to which the current vehicle belongs according to the time tag of the current track point; in the embodiment, the total time is divided into a plurality of time intervals by taking 15min as a time interval, and the traffic flow in each time interval is counted, as shown in fig. 2 and 3;
step 1.1.4, accumulating the current vehicle to the traffic flow matrix (mu)itλ), i represents a traveling direction, i ∈ [1,12]](ii) a t denotes the number of the time interval, μitA traffic flow representing i a travel direction t time interval; λ is intersection identification, e.g. (μ)it1) a traffic flow matrix representing intersection 1;
step 1.2, dividing high and low peak time periods according to the traffic flow corresponding to each time interval; the average value of the traffic flow in a plurality of periods can be calculated and used as the judgment standard of the high and low peak of the traffic flow to divide the high and low peak periods, the average value of the traffic in a week is calculated in the embodiment and used as the judgment standard of the high and low peak of the traffic flow, and the time interval of each day is divided into the high and low peak periods;
after the traffic flows in different advancing directions of the intersection are calculated, a mathematical model for solving high, medium and low flow thresholds is constructed on the basis of high and low peak time interval division. Considering the difference between historical data and real-time traffic data, which may cause errors in the selection of peak time, in order to reflect the dynamic performance of the model, the model is designed to set dynamic weight parameters s for different peak time intervals according to the real-time traffic flowjDefining weight formula sj
sj＝α/γ
Wherein, alpha is the traffic flow of the intersection at the current peak time (a time interval), and gamma is the sum of the traffic flows at all the peak time of the current intersection;
then establishing the traffic flow of the single intersection in the peak period:
wherein, j ∈ (peak time), i ∈ [1,12]]I.e. the direction of travel of the vehicle, num denotes the number of peak hours, μijIs an element in the traffic matrix above; thus established by the above definition: tuple representation (or vector representation) of high and low peak period traffic of a single intersection, and peak period traffic tuple D of a single intersectionthIs composed of
Dth＝[dth1,dth2,...,dth12]
Similar to the above method, the peak-low time interval traffic tuple D of a single intersection can be obtainedtlIs composed of
Dtl＝[dtl1,dtl2,...,dtl12]
Step 1.3, ordering the traffic flow in the same direction of a single intersection, wherein the flow matrix of each intersection is
Wherein q isIJThe traffic flow value after the traffic flow sorting is shown, wherein I is 12, 1 to I represent the traveling direction, J represents the time interval, and the time interval is the time interval, in the embodiment, 6 to 23 points of each day are taken as one period, the time of one period is divided into 72 time intervals according to the time interval of 15min, and J ∈ [1,72 ]]；
Step 1.4 according to DthAnd DtlD in the same direction of travelthiAnd dtliDividing time intervals in the same traveling direction in the sequenced flow matrix Q into low-flow areas S corresponding to the traffic flowLMedium flow rate region SMIs high and highFlow area SH：
The lengths of the three sequences are respectively marked as l1, l2 and l 3; for the i traveling direction, respectively take SL，SMIs used as the middle and low flow representative value r in the i traveling directionliAnd the representative value r of the sum flowmiTaking SH70 decimals of (a) as a high flow rate representative value r in the ith directionhi：
Representing the flow threshold matrix as R;
step 2, selecting traffic information characteristics for dividing the influence control subarea, wherein the traffic information characteristics comprise: traffic flow influence characteristics, intersection distance influence characteristics, traffic flow composition influence characteristics and cycle influence characteristics;
(1) influence analysis model
When the traffic flow of the road junction is too small, the coordination control function is not obvious, and a traffic flow influence coeffient (TF) is established:
wherein q (t) is the flow of t time interval, sigma is the flow of one direction of t time interval of the intersection, and lambda is the weight factor of different traffic flow directions; the introduction of lambda actually adjusts the calculated TF value to be between 0 and 100;
(2) model of influence of intersection distance
The influence model of the intersection distance is an important static factor of the division of an adjacent intersection which is a traffic control subarea, the common division principle adopts the principle of maximum and minimum intersection distance, and the distance between the adjacent intersections is set to be L;
the maximum distance principle is as follows: when the upstream vehicle drives to the downstream intersection, the discrete type enhancement is caused along with the increase of the intersection distance, so that the coordination control requirement among the intersections is smaller and smaller. Traffic engineering experience indicates that the intersection distance length for distinguishing whether the vehicle discreteness is within the coordination control range is 800 m.
Spacing minimum principle: the crossing dredging capacity at any moment is required to be capable of releasing all vehicles which flow to the current crossing, otherwise, vehicle queuing and secondary queuing cause traffic jam.
The division basis of the traffic control subareas in the existing maximum and minimum distance principle is obtained only according to traffic experience and is defined as static parameters, and the dynamic traffic characteristics of the current road network are ignored, so that the principle loses generality in the process of dividing the control subareas aiming at the road networks of different types and different time periods. The maximum-minimum distance principle based on dynamic flow change can make up for the defect, and whether vehicles introduced into the adjacent intersection are successfully exported and the discrete type between the adjacent vehicles is kept is an important basis for defining the division rule of the length of the road section because the length of the road section determines the traffic capacity of the adjacent intersection.
The improved maximum-minimum spacing principle of the invention is as follows:
maximum spacing principle:
ρ＝avg(C1(t),C2(t))
in the maximum distance principle, RminAverage value of high flow representative values in two adjacent crossroads, NinThe direction of a vehicle which can run in the current green light period; c1(t)、C2(t) respectively representing the signal periods of a front intersection and a back intersection in two adjacent intersections; t issampleIs the length of a time unit, ρ is the average value in the signal period of an adjacent intersection, LmaxRepresents the maximum distance between adjacent intersections, hcarIs the average length of the vehicle, NlaneIs the number of lanes, β vehicle discrete distance, F traffic flow discrete coefficient;
minimum spacing principle:
ρ＝max(C1(t),C2(t))
in the principle of minimum spacing, RmaxLarger value of high flow rate representative value, T, in two adjacent crossroadssampleIs the length of a time unit, ρ is the larger value in the signal period of the adjacent intersection, μ is the anti-congestion factor, LminIndicates the minimum distance between adjacent intersections, hcarIs the vehicle average length;
on the basis of the intersection maximum-minimum principle, an intersection distance influence characteristic IF (intersection distance influence) is established:
wherein χ is a balance factor; l is the length of the current road section;
(3) traffic flow composition influence coefficient model
If no vehicles come between two intersections, no coordination control requirement exists between the two intersections, otherwise, if the vehicles coming from the two intersections are the main constituent parts of the traffic flow of each other, traffic jam can not be formed only when the two intersections are subjected to coordination control. Therefore, the traffic flow composition is an important parameter of traffic information characteristics, the discrete magnitude of the traffic flow and the coordination magnitude of the upstream and downstream intersections are determined for the proportion of the vehicles at the upstream intersection in the straight-ahead vehicles at the downstream intersection, and the traffic flow composition influence characteristics ci (compositional information):
CI＝O*F
wherein O is the proportion of the vehicles which go straight upstream to continue to go straight downstream, the value range [0,1] is the traffic flow discrete coefficient, and F is the traffic flow discrete coefficient;
(4) periodic influence model
The periodic influence model is characterized in that when the period length difference between adjacent intersections is too large and an integer proportion cannot be formed, the necessity of coordinating the intersections is reduced, and a periodic influence characteristic CE (cycle effect) is established:
wherein, T1The long green time, T, of the green time in two adjacent crossroads2The green light time is the short green light time in the green light time in two adjacent intersections;
step 3, establishing an influence weight calculation model based on space-time characteristics to obtain a weight L 'of the characteristics r'r：
Step 3.1, all samples of the characteristic r in the traffic information characteristic are taken to form a set XrI.e. the set obtained for all the different time intervals of a feature:
Xr＝[x1,x2,...,xn]T
where n is the total number of samples, r ∈ [1,4 ]]The characteristic labels in the traffic information characteristics represent traffic flow influence characteristics, intersection distance influence characteristics, traffic flow composition influence characteristics and cycle influence characteristics; x is the number of1,x2,...,xnSample representation forms of characteristic values corresponding to the characteristic r respectively;
if a sample point xi'Is another sample point xj'Is usually k-5, then x isi'And xj'An edge exists between the two, and the rule is utilized to respectively establish a feature undirected graph of each feature;
step 3.2, calculating the weight W of each edge in the characteristic undirected graphi'j'Form a weight matrix WrForming an adjacent matrix;
step 3.3, calculating the variance of all samples of the characteristics; drIs a diagonal matrix whose diagonal element value is the sum of the out-of-measure weights of each sample point and is based on DrThe element value in (1) is an intermediate variable X'r：
Dr＝diag(Wr*1) 1＝(1,1,...,1)T
The out-of-measure weight sum is a complete noun, and the calculated sum is the weight sum of all edges starting from a node;
thereby calculating the variance of all samples of the feature
Step 3.4, knowing WrThe weight matrix is used for carrying out influence weight calculation on the space-time characteristics on the characteristics, and then the calculation can obtain
Lr＝Dr-Wr
xri'、xrj'Respectively representing sample points x corresponding to the features ri'、xj'；
The weight of the feature r is therefore:
step 4, a weighted community discovery algorithm (a Louvain algorithm of dynamic modularity division) of dynamic modularity division:
the method optimizes the Louvain algorithm, improves the phenomenon that iteration times are excessive in the process of merging communities, does not really calculate a community structure which can maximize the modularity value in the process of merging calculation, controls the iteration times by defining the threshold value e, and can determine that the community structure of the weighted directed graph tends to be stable when the true value of delta Q is smaller than e without carrying out community merging.
Under the condition of the known modularity definition, the weighted community discovery algorithm for dynamic modularity division realizes the division of traffic control subareas, and the specific process comprises the following steps:
step 4.1, obtaining a calculation coordination coefficient CC between every two intersections by performing product accumulation on each feature weight calculated by the influence weight calculation model based on the space-time featuresij(coding coefficient) as the weight of the road section, wherein R is the values of IF, CI, CE and TF respectively;
step 4.2, when initializing a community matrix, each node in the graph is an independent community, the number of communities in the initial state is consistent with the number of nodes, the adjacent matrix represents the weighted undirected graph, and the degree weight sum of each node is calculated;
4.3, each node and adjacent nodes try to carry out combined community operation to obtain a modularity variation value delta Q before and after distribution, record a community combination state with the maximum delta Q value and the delta Q value larger than 0, and distribute the nodes to communities where the adjacent nodes are located, otherwise, the nodes are kept unchanged;
4.4, completing one merging operation until the modularity change value caused by the community structure change of the weighted undirected graph is smaller than a threshold value, namely the modularity value is maximum;
step 4.5, compressing the graph according to the recorded combined community state, reconstructing all nodes in the same community into a new node, creating a new community matrix, converting the weight of edges among the nodes in the community into the weight of a new node ring, and converting the weight of edges among the communities into the weight of edges among the new nodes;
and 4.6, repeating the calculation process from the step 4.2 to the step 4.5 until the modularity variation value of the whole graph is smaller than the threshold value, and finally realizing the division of the traffic control subareas.
Examples
1. Data pre-processing
The data set used by the invention is GPS records of 1.4 ten thousand taxis in metropolis, the time is from 8, month 3 to 8, month 30 in 2014, the data of the time period from zero to six morning are ignored, repeated and abnormal records are cleaned, each record comprises a taxi ID, a latitude, a longitude and a timestamp, and the time interval of each record is 60 s. The data is in the form:
1，30.624806，104.136604，1，2014/8/3 21:18:46
1，30.624809，104.136612，1，2014/8/3 21:18:15
1，30.624811，104.136587，1，2014/8/3 21:20:17
1，30.624811，104.136596，1，2014/8/3 21:19:16
when a large number of track points are displayed in chronological order, the traffic pattern, that is, the shape of the road network, can be displayed. Therefore, in the experiment, the longitude and latitude are mapped into the plane coordinates, the coordinates are connected according to the time stamps to form the vehicle track, the road network is displayed by displaying the route map of the real track, and the shapes of the track road network and the real map can be compared as shown in fig. 4 and fig. 5. The method has the advantages that the time sequence display of the track points is not greatly different from that of a real road network, the accuracy and the authenticity of the track point data are proved, the value and the reliability of experiments based on the track points are ensured, and the data are verified to have no large data noise.
Because the research focuses on the similarity of the intersections, the types of the intersections must be unified, and the types of the intersections researched by the invention are unified into the intersections. The number of crossroads, the geographic coordinates of the crossroads, the number of directions in which the crossroads can pass and the connectivity among the crossroads are all characteristic points to be determined. The method comprises the steps of selecting a clear and obvious area of the crossroad, obtaining the real coordinate of the crossroad by using a ginput function in matlab, and selecting 35 crossroads for research, as shown in fig. 6.
2. Results and analysis of the experiments
According to the determined intersection region range, counting the number of vehicles coming in and going out of the region within 15 minutes, determining the peak time periods of the traffic flow by the average value of the vehicle flow of each time period, and calculating the high, medium and low flow threshold values of the traffic flow, so that the dynamic flow threshold value of each intersection can be obtained, and the threshold value is synchronous with the traffic flow change. The high, medium and low flow threshold values are used for determining the judgment coefficient of the model, and then the coordination coefficient of each intersection based on the traffic flow characteristics is dynamically calculated. Each intersection has 12 different traffic flow advancing directions, and for one intersection, the traffic flow data of each advancing direction of the same intersection are counted by using a traffic flow calculation model with 15min as a time sampling interval. And substituting the influence factors into a traffic flow influence model to calculate the influence factor based on the flow of each intersection, and calculating the road section length and the traffic flow composition ratio of the road network by using each model in the same way to calculate the influence factor of each traffic characteristic.
The influence degree of each traffic characteristic in the area is different, and the weight coefficient of each characteristic element needs to be determined according to data association in each characteristic element through an influence weight calculation model. The method determines four traffic characteristics of traffic flow, road section length, traffic composition and signal period, wherein 3816 sample elements are totally arranged in each characteristic and are substituted into an influence weight calculation model to obtain the weight coefficient of the traffic characteristics, as shown in fig. 7.
In the figure, the weight numbers of the traffic flow, the link length, the traffic cycle and the traffic flow calculated by the spatio-temporal feature influence weight calculation model are respectively 30.37%, 13.46%, 22.88% and 33.29%.
And combining the traffic characteristics with the weight coefficients, then calculating to obtain characteristic values serving as the weight of each road section, substituting the characteristic values into a Louvain algorithm of dynamic modular degree division, and fusing the topological structure characteristics of the road network into the process of dividing the traffic control subareas. The road network is represented in the form of an adjacency list and is used as input data of a dynamic modularity dividing Louvain algorithm, and the road network is divided into communities continuously until the change value of the modularity is smaller than a threshold value e, namely the community combination of the road network has no influence on the modularity. The calculated control sub-division state is shown in fig. 8.
The 35 crossroads are divided into 5 traffic control sub-areas, wherein the intersections within the same shaded area belong to the same traffic control sub-area, i.e., (1, 2, 3, 4, 5, 13, 14, 15), (6, 7, 8, 19, 20, 21, 27, 28), (9, 10, 11, 12, 22, 23, 24, 33, 34, 35), (16, 17, 18, 25, 26, 29, 30), (31, 32). Each control subarea needs to be subjected to independent signal coordination control, and two community combinations are performed in the community division process, wherein the combination state is shown in fig. 9.
3. Simulation and verification
The VISSIM simulation software is a miniature simulation modeling tool based on time sections and driving behaviors and is used for realizing the modeling of urban and public transport operation. The method analyzes the running conditions of cities and public transport under various traffic condition combination states, such as lane quantitative index setting, traffic composition parameter adjustment, traffic signal control adjustment and the like, and is a simulation tool for evaluating the realizability of engineering design and the effectiveness of urban planning. The software internally comprises a traffic simulator and a signal state generator, and the sharing of the detector data and the signal state information is realized through an interface. The VISSIM can output various statistics offline such as travel time, queue length, etc.
The invention adopts VISSIM4.3 traffic microscopic simulation software to verify the mathematical model of the invention for dividing the control subareas. As shown in fig. 10, VISSIM simulation software is run on the Windows 7 operating system, and the real network is imported into the software.
The length of each road section is determined by the length of a real road section, all roads in the experiment are provided with two lanes, the traffic cycle of each intersection is input according to a traffic cycle model, traffic signals in the experiment adopt a two-phase form, the signal cycle is 120s, and a signal timing scheme of each intersection is calculated by a dynamic traffic cycle model.
And inputting the acquired traffic flow state into VISSIM simulation software, setting a travel in a road network, recording the time length of each vehicle entering and leaving the road network, and setting an acquisition target of a test document, such as data acquisition, queue length, travel time, delay time and the like.
According to the test document, the running state of the vehicle and the traffic running condition under the original traffic control state can be obtained. And importing a control subarea division model obtained by calculation of a Louvain algorithm divided by dynamic modules into VISSIM simulation software, unifying traffic periods at crossroads of the same control subarea, and performing simulation under the condition that other traffic characteristics and test environments are not changed. In the experiment, the simulation time length is consistent with the time length of the time unit and is 15 minutes, namely 900 s. In the experiment, 5 traffic control sub-areas which are required to be divided in the road network are calculated by a dynamic modularity Louvain algorithm, and a signal timing scheme of each control sub-area calculated by a dynamic traffic periodic model is shown in FIG. 11, wherein the periodic timing schemes of the same traffic control sub-area are the same. In fig. 11, the total Time of the traffic light Cycle is defined by Cycle Time, the yellow light Time is set by Amber, the green light end Time is set by GreenEnd, the red light end Time is set by RedEnd, and the Time Cycle of the two-phase traffic light is completely set.
After the simulation operation is finished, four files of rsr, vlr, mes and stz are generated in total, and the travel time, the delay time, the data acquisition and the queuing length are recorded respectively. In the experiment, 5 travel routes are selected, all the travels cover all the dense vehicle road sections and different traffic control sub-areas, and the comparison is performed from two aspects of average travel time and vehicle driving information for comparing the effectiveness of dividing the mathematical model for the control sub-areas.
TABLE 1 travel time comparison
As shown in Table 1, the traffic control subarea division scheme optimizes the travel time of 5 trips to different degrees, and the longer the distance of the trips is, the more obvious the optimization effect is.
TABLE 2 vehicle Integrated Driving parameter comparison
As shown in Table 2, in the traveling process, the average queuing length of each course is reduced by 30%, the average delay time is shortened by 44.6%, the average traveling speed is slightly increased, the speed is increased by 7.2%, and the average stopping frequency is reduced by 30%. The scheme for dividing the traffic control subareas is improved on 4 driving parameters, and the travel experience of people is improved.
The traffic control subarea division mathematical model provided by the invention has a good effect on optimizing the vehicle travel, effectively reduces the travel time of the vehicle in long-distance travel, shortens the delay time of the vehicle in travel, increases the saturation and the utilization rate of roads, effectively reduces the parking times, improves the travel speed to a certain extent and comprehensively optimizes the travel efficiency and road conditions under the same travel distance.
Claims (9)
1. The self-adaptive traffic control subarea division method based on the space data mining is characterized by comprising the following steps of:
step 1, extracting characteristics of high and low peak time periods and high, medium and low flow thresholds;
step 2, selecting traffic information characteristics influencing division of the control subarea;
the traffic information features include: traffic flow influence characteristic TF, intersection distance influence characteristic IF, traffic flow composition influence characteristic CI and period influence characteristic CE;
step 3, establishing an influence weight calculation model based on space-time characteristics to obtain a weight L 'of the characteristics r'r，r∈[1,4]The characteristic labels in the traffic information characteristics represent traffic flow influence characteristics TF, intersection distance influence characteristics IF, traffic flow composition influence characteristics CI and cycle influence characteristics CE;
step 4, realizing the division of the traffic control subareas by a weighted community discovery algorithm based on dynamic modularity division, wherein the specific process comprises the following steps:
step 4.1, obtaining a calculation coordination coefficient CC between every two intersections by performing product accumulation on each feature weight calculated by the influence weight calculation model based on the space-time featuresijAs the weight of the road section, wherein R is the values of IF, CI, CE and TF respectively;
step 4.2, when initializing a community matrix, each node in the graph is an independent community, the number of communities in the initial state is consistent with the number of nodes, the adjacent matrix represents the weighted undirected graph, and the degree weight sum of each node is calculated;
4.3, each node and adjacent nodes try to carry out combined community operation to obtain a modularity variation value delta Q before and after distribution, record a community combination state with the maximum delta Q value and the delta Q value larger than 0, and distribute the nodes to communities where the adjacent nodes are located, otherwise, the nodes are kept unchanged;
4.4, completing one merging operation until the modularity change value caused by the community structure change of the weighted undirected graph is smaller than a threshold value, namely the modularity value is maximum;
step 4.5, compressing the graph according to the recorded combined community state, reconstructing all nodes in the same community into a new node, creating a new community matrix, converting the weight of edges among the nodes in the community into the weight of a new node ring, and converting the weight of edges among the communities into the weight of edges among the new nodes;
and 4.6, repeating the calculation process from the step 4.2 to the step 4.5 until the modularity variation value of the whole graph is smaller than the threshold value, and finally realizing the division of the traffic control subareas.
2. The method for partitioning the sub-area of the adaptive traffic control based on the spatial data mining as claimed in claim 1, wherein the step 1 of performing the feature extraction of the high and low peak periods and the high, medium and low flow thresholds comprises the following steps:
step 1.1, calculating traffic flows in different traveling directions passing through the intersection, namely calculating the traffic flows of roads which enter the intersection from one of the two intersected roads and then travel to different directions;
step 1.2, dividing high and low peak time periods according to the traffic flow corresponding to each time interval:
after the traffic flows in different advancing directions of the intersection are calculated, a mathematical model for solving high, medium and low flow thresholds is constructed on the basis of high and low peak time interval division; the design sets dynamic weight parameters s for different high and low peak time intervals according to real-time traffic flowjDefining weight formula sj
sj＝α/γ
Wherein alpha is the traffic flow of the intersection at the current peak time, and gamma is the sum of the traffic flows of all the peak time of the intersection;
then establishing the traffic flow of the single intersection in the peak period:
of these, j ∈ (high)Peak period), i ∈ [1,12]I.e. the direction of travel of the vehicle, num represents the number of peak hours; mu.sijA traffic flow representing i a travel direction j time interval;
establishing: tuple representation of high and low peak time interval flow of single intersection, and high peak time interval flow tuple D of single intersectionthIs composed of
Dth＝[dth1,dth2,...,dth12]
Similar to the above method, the peak-low time interval traffic tuple D of a single intersection can be obtainedtlIs composed of
Dtl＝[dtl1,dtl2,...,dtl12]
Step 1.3, ordering the traffic flow in the same direction of a single intersection, wherein the flow matrix of each intersection is
Wherein q isIJThe traffic flow value after the traffic flow sorting is shown, wherein I is 12, 1 to I represent the traveling direction, J represents a time interval, and the time interval is marked as a time interval;
step 1.4 according to DthAnd DtlD in the same direction of travelthiAnd dtliDividing time intervals in the same traveling direction in the sequenced flow matrix Q into low-flow areas S corresponding to the traffic flowLMedium flow rate region SMHigh flow area SH：
The lengths of the three sequences are respectively marked as l1, l2 and l 3; for the i traveling direction, respectively take SL，SMIs used as the middle and low flow representative value r in the i traveling directionliAnd the representative value r of the sum flowmiTaking SH70 decimals of (a) as a high flow rate representative value r in the ith directionhi：
Representing the flow threshold matrix as R;
3. the method for partitioning the sub-area of the adaptive traffic control based on the spatial data mining as claimed in claim 2, wherein the specific process of the step 1.1 comprises the following steps:
step 1.1.1, marking two crossed roads of the intersection as a first road and a second road respectively, and taking two side edges of the first road and the second road as boundaries; the intersection divides two intersecting roads into 5 areas which are respectively an intersection, a front section of a first road intersection, a rear section of the first road intersection, a front section of a second road intersection and a rear section of the second road intersection;
step 1.1.2, judging the original driving direction of the vehicle, namely the direction from which the vehicle enters the intersection;
then, judging the traveling direction of the vehicle according to the subsequent track direction of the current vehicle;
step 1.1.3, judging a time interval to which the current vehicle belongs according to the time tag of the current track point;
step 1.1.4, accumulating the current vehicle to the traffic flow matrix (mu)itλ), i represents a traveling direction, i ∈ [1,12]](ii) a t denotes the number of the time interval, μitA traffic flow representing i a travel direction t time interval; and lambda is the intersection mark.
4. The method for dividing the sub-area of the adaptive traffic control based on the spatial data mining according to claim 3, wherein the traffic flow influence characteristics in the step 2 are as follows:
traffic flow influence characteristic TF:
wherein q (t) is the flow of t time interval, sigma is the flow of one direction of t time interval of the intersection, and lambda is the weight factor of different traffic flow directions; the introduction of λ effectively adjusts the calculated TF value to between 0 and 100.
5. The method for partitioning the sub-area of the adaptive traffic control based on the spatial data mining as claimed in claim 4, wherein the intersection distance influence characteristics in the step 2 are as follows:
setting the distance between adjacent intersections as L;
the maximum-minimum spacing principle is as follows:
maximum spacing principle:
ρ＝avg(C1(t),C2(t))
in the maximum distance principle, RminAverage value of high flow representative values in two adjacent crossroads, NinThe direction of a vehicle which can run in the current green light period; c1(t)、C2(t) respectively representing the signal periods of a front intersection and a back intersection in two adjacent intersections; t issampleIs the length of a time unit, ρ is the average value in the signal period of an adjacent intersection, LmaxRepresents the maximum distance between adjacent intersections, hcarIs the average length of the vehicle, NlaneIs the number of lanes, β vehicle discrete distance, F traffic flow discrete coefficient;
minimum spacing principle:
ρ＝max(C1(t),C2(t))
in the principle of minimum spacing, RmaxLarger value of high flow rate representative value, T, in two adjacent crossroadssampleIs the length of a time unit, ρ is the larger value in the signal period of the adjacent intersection, μ is the anti-congestion factor, LminIndicates the minimum distance between adjacent intersections, hcarIs the vehicle average length;
on the basis of the maximum-minimum principle of the intersection, establishing intersection distance influence characteristics IF:
wherein χ is a balance factor; and l is the length of the current road section.
6. The method for dividing the sub-area of the adaptive traffic control based on the spatial data mining according to claim 1, wherein the traffic flow composition influence characteristics in the step 2 are as follows:
the traffic flow composition is an important parameter of traffic information characteristics, the discrete size of traffic flow and the coordination size of upstream and downstream intersections are determined for the proportion of upstream intersection vehicles in the downstream intersection straight vehicles, and the traffic flow composition influence characteristic CI is established:
CI＝O*F
wherein, O is the proportion of the vehicles which go straight at the upstream and continue to go straight at the downstream, the value range [0,1], and F is a traffic flow discrete coefficient.
7. The method for partitioning the sub-area of the adaptive traffic control based on the spatial data mining as claimed in claim 1, wherein the periodic impact characteristics in step 2 are as follows:
establishing a periodic impact characteristic CE:
wherein, T1The long green time, T, of the green time in two adjacent crossroads2The green time is the short one of the green time in two adjacent crossroads.
8. The method for partitioning sub-area of adaptive traffic control based on spatial data mining according to one of claims 1 to 7, wherein the step 3 of establishing the model for calculating influence weight based on space-time characteristics obtains the weight L 'of the characteristic r'rThe process of (2) is as follows:
step 3.1, all samples of the characteristic r in the traffic information characteristic are taken to form a set Xr：
Xr＝[x1,x2,...,xn]T
Where n is the total number of samples, r ∈ [1,4 ]]The characteristic labels in the traffic information characteristics represent traffic flow influence characteristics, intersection distance influence characteristics, traffic flow composition influence characteristics and cycle influence characteristics; x is the number of1,x2,...,xnSample representation forms of characteristic values corresponding to the characteristic r respectively;
if a sample point xi'Is another sample point xj'Is one of the nearest k sample points, then xi'And xj'An edge exists between the two, and the rule is utilized to respectively establish a feature undirected graph of each feature;
step 3.2, calculating the weight W of each edge in the characteristic undirected graphi'j'Form a weight matrix WrForming an adjacent matrix;
step 3.3, calculating the variance of all samples of the characteristics; drIs a diagonal matrix whose diagonal element value is the sum of the out-of-measure weights of each sample point and is based on DrThe value of the element in (1) yields the intermediate variable Xr′：
Dr＝diag(Wr*1)1＝(1,1,...,1)T
Thereby calculating the variance of all samples of the feature
Step 3.4, knowing WrThe weight matrix is used for carrying out influence weight calculation on the space-time characteristics on the characteristics, and then the calculation can obtain
Lr＝Dr-Wr
xri'、xrj'Respectively representing sample points x corresponding to the features ri'、xj'；
The weight of the feature r is therefore:
9. the adaptive traffic control subdivision method based on spatial data mining of claim 8, wherein k is 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811332450.5A CN109410577B (en) | 2018-11-09 | 2018-11-09 | Self-adaptive traffic control subarea division method based on space data mining |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811332450.5A CN109410577B (en) | 2018-11-09 | 2018-11-09 | Self-adaptive traffic control subarea division method based on space data mining |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109410577A CN109410577A (en) | 2019-03-01 |
| CN109410577B true CN109410577B (en) | 2020-10-09 |
Family
ID=65472438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811332450.5A Active CN109410577B (en) | 2018-11-09 | 2018-11-09 | Self-adaptive traffic control subarea division method based on space data mining |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109410577B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110033613B (en) * | 2019-03-07 | 2020-07-10 | 吉林建筑大学 | Smart city management method and system based on regional traffic synchronism |
| CN109934850B (en) * | 2019-03-21 | 2021-04-30 | 北京沃东天骏信息技术有限公司 | Method, device and system for counting moving objects |
| CN110111567B (en) * | 2019-04-23 | 2021-05-18 | 刘畅 | Traffic control subarea division method and system based on modularity evaluation |
| CN112187499B (en) * | 2019-07-03 | 2021-12-03 | 四川大学 | Device partition management and division method in device network |
| CN110415523B (en) * | 2019-08-13 | 2021-07-30 | 东南大学 | Signal control subarea division method based on vehicle travel track data |
| CN110634287B (en) | 2019-08-26 | 2021-08-17 | 上海电科智能系统股份有限公司 | Urban traffic state refined discrimination method based on edge calculation |
| CN111145548B (en) * | 2019-12-27 | 2021-06-01 | 银江股份有限公司 | Important intersection identification and subregion division method based on data field and node compression |
| CN111210625B (en) * | 2020-01-10 | 2022-06-07 | 阿里巴巴集团控股有限公司 | Traffic control method and device and electronic equipment |
| CN113129614B (en) * | 2020-01-10 | 2023-01-24 | 阿里巴巴集团控股有限公司 | Traffic control method and device and electronic equipment |
| CN111275964B (en) * | 2020-01-14 | 2021-03-23 | 浙江浙大中控信息技术有限公司 | Road section correlation matrix calculation method based on checkpoint data |
| CN111325979B (en) * | 2020-02-28 | 2021-07-16 | 海信集团有限公司 | Method and device for dividing traffic control multistage subareas |
| CN111192457A (en) * | 2020-03-20 | 2020-05-22 | 青岛海信网络科技股份有限公司 | Method and device for controlling urban-level integral traffic signals |
| CN114973699A (en) * | 2022-05-10 | 2022-08-30 | 阿波罗智联(北京)科技有限公司 | Traffic control signal generation method, edge calculation unit and road side unit |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002159048A (en) * | 2000-11-22 | 2002-05-31 | Yrp Mobile Telecommunications Key Tech Res Lab Co Ltd | Cdma mobile communication system |
| KR20050051956A (en) * | 2003-11-28 | 2005-06-02 | 주식회사 비츠로시스 | Control system and methdod for local divisional traffic signal |
| CN101751777A (en) * | 2008-12-02 | 2010-06-23 | 同济大学 | Dynamic urban road network traffic zone partitioning method based on space cluster analysis |
| CN105825690A (en) * | 2016-06-15 | 2016-08-03 | 北京航空航天大学 | Coordinated control oriented trunk line crossing correlation analysis and division method |
| CN105869401A (en) * | 2016-05-12 | 2016-08-17 | 华南理工大学 | Road network dynamic partitioning method based on different crowdedness degrees |
| CN108320511A (en) * | 2018-03-30 | 2018-07-24 | 江苏智通交通科技有限公司 | Urban highway traffic sub-area division method based on spectral clustering |
| CN108388961A (en) * | 2018-02-06 | 2018-08-10 | 华东师范大学 | Self-adapting random neighbours' community detecting algorithm based on modularity optimization |
-
2018
- 2018-11-09 CN CN201811332450.5A patent/CN109410577B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002159048A (en) * | 2000-11-22 | 2002-05-31 | Yrp Mobile Telecommunications Key Tech Res Lab Co Ltd | Cdma mobile communication system |
| KR20050051956A (en) * | 2003-11-28 | 2005-06-02 | 주식회사 비츠로시스 | Control system and methdod for local divisional traffic signal |
| CN101751777A (en) * | 2008-12-02 | 2010-06-23 | 同济大学 | Dynamic urban road network traffic zone partitioning method based on space cluster analysis |
| CN105869401A (en) * | 2016-05-12 | 2016-08-17 | 华南理工大学 | Road network dynamic partitioning method based on different crowdedness degrees |
| CN105825690A (en) * | 2016-06-15 | 2016-08-03 | 北京航空航天大学 | Coordinated control oriented trunk line crossing correlation analysis and division method |
| CN108388961A (en) * | 2018-02-06 | 2018-08-10 | 华东师范大学 | Self-adapting random neighbours' community detecting algorithm based on modularity optimization |
| CN108320511A (en) * | 2018-03-30 | 2018-07-24 | 江苏智通交通科技有限公司 | Urban highway traffic sub-area division method based on spectral clustering |
Non-Patent Citations (5)
| Title |
|---|
| 交通信号控制子区模糊动态划分方法研究;李瑞敏等;《武汉理工大学学报》;20080630;第32卷(第3期);381-384 * |
| 交通小区划分问题的整数规划建模与优化算法研究;王霖青;《中国博士学位论文全文数据库 工程科技Ⅱ辑》;20161215(第12期);C034-16 * |
| 基于交通流预测的控制子区交通状态识别技术研究;朱芸;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20140815(第8期);C034-180 * |
| 短时交通状态预测下交通控制子区自动划分方法;郭海锋等;《系统科学与数学》;20150831;第35卷(第8期);904-918 * |
| 面向交通控制的时段划分与子区划分;赵伟明;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20140615(第6期);C034-624 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109410577A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109410577B (en) | Self-adaptive traffic control subarea division method based on space data mining | |
| CN111612206B (en) | Neighborhood people stream prediction method and system based on space-time diagram convolution neural network | |
| CN109448370B (en) | Traffic control subarea division method based on vehicle track data | |
| US10354523B2 (en) | Road traffic control system, method, and electronic device | |
| CN112700072B (en) | Traffic condition prediction method, electronic device, and storage medium | |
| CN111583639B (en) | Road traffic jam early warning method and system | |
| KR101638368B1 (en) | Prediction System And Method of Urban Traffic Flow Using Multifactor Pattern Recognition Model | |
| CN105513370B (en) | The traffic zone division methods excavated based on sparse license plate identification data | |
| CN110222873A (en) | A kind of subway station passenger flow forecast method based on big data | |
| EP2590151A1 (en) | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras | |
| CN105489006B (en) | Multiple dimensioned vehicle flow visual analysis method based on GPS data from taxi | |
| CN112466117A (en) | Road network short-term traffic flow prediction method based on deep space-time residual error network | |
| CN108597227A (en) | Road traffic flow forecasting method under freeway toll station | |
| CN108986453A (en) | A kind of traffic movement prediction method based on contextual information, system and device | |
| CN106157624B (en) | More granularity roads based on traffic location data shunt visual analysis method | |
| CN110021161B (en) | Traffic flow direction prediction method and system | |
| WO2022142418A1 (en) | Traffic performance index prediction method and device based on gis map information | |
| CN106815563A (en) | A kind of crowd's quantitative forecasting technique based on human body apparent structure | |
| CN110032952A (en) | A kind of road boundary point detecting method based on deep learning | |
| CN113868492A (en) | Visual OD (origin-destination) analysis method based on electric police and checkpoint data and application | |
| CN112863182A (en) | Cross-modal data prediction method based on transfer learning | |
| CN114372503A (en) | Cluster vehicle motion trail prediction method | |
| CN110070720B (en) | Calculation method for improving fitting degree of traffic capacity model of intersection road occupation construction area | |
| Prontri et al. | Traffic signal control using fuzzy logic | |
| CN108053646A (en) | Traffic characteristic acquisition methods, Forecasting Methodology and system based on time-sensitive feature |
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 |