CN114170796B - Algorithm improved congestion propagation analysis method - Google Patents
Algorithm improved congestion propagation analysis method Download PDFInfo
- Publication number
- CN114170796B CN114170796B CN202111381097.1A CN202111381097A CN114170796B CN 114170796 B CN114170796 B CN 114170796B CN 202111381097 A CN202111381097 A CN 202111381097A CN 114170796 B CN114170796 B CN 114170796B
- Authority
- CN
- China
- Prior art keywords
- intersection
- item
- congestion
- traffic
- lane
- 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
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- 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
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- 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
- G08G1/0133—Traffic data processing for classifying traffic situation
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to the field of congestion propagation analysis methods, in particular to an algorithm improved congestion propagation analysis method. The method comprises the following steps: a) The congestion judging module processes data from the region, divides the traffic state by calculating the intersection saturation, and calculates the saturation of each lane of the intersection by calculating the traffic demand index TI and the intersection traffic capacity so as to realize the division of the traffic state; b) The item set generation module integrates the results obtained from the congestion judgment module; c) And generating an FP-Tree frequent item set. The frequent item set mining method of the Apriori algorithm is replaced by the FP-Tree algorithm, and the operation efficiency of the algorithm is effectively improved. And completing data filtering by adding space-time characteristic constraints, and adding a promotion degree evaluation method to filter invalid data in the strong association rule again. The method and the device can rapidly analyze the congestion tendency and provide information for city congestion relief.
Description
Technical Field
The invention relates to the field of congestion propagation analysis methods, in particular to an algorithm improved congestion propagation analysis method.
Background
The problem of urban traffic congestion is a serious obstacle to urban traffic development, so that the road traffic efficiency is reduced, the road traffic safety is seriously influenced, and the happiness of citizens is also reduced. The problem of traffic jam is not only related to traffic flow and road facilities, but also inseparable from citizen travel rules and signal timing, and the analysis of the cause of the traffic jam phenomenon is very important to delay jam. Meanwhile, the urban road network is provided with a network propagation structure in space, traffic jam has strong propagation in the road network, especially in early and late rush hours, traffic jam is easy to generate at bottleneck intersections, if control measures are not taken, the jam is likely to propagate upstream and downstream through road sections connected with the bottleneck intersections, and more intersections may be impacted to cause larger-scale jam.
In order to alleviate the problem of spreading urban traffic congestion, scholars at home and abroad begin to gradually study the spreading phenomenon of traffic congestion, and various methods based on spreading dynamics, infectious disease SIS models and cellular automata are developed, but the method does not have good space-time characteristics for spreading congestion. Although the Apriori algorithm based on data association rule mining can highlight the space-time characteristics, the data scale is proportional to the number of intersections in the road network, and the larger the number of intersections included in the road network is, the larger the data volume is. The conventional Apriori algorithm cannot filter invalid data well, and the operation efficiency of the algorithm is too low.
Disclosure of Invention
In order to solve the technical problems described in the background art, the invention provides an algorithm improved congestion propagation analysis method. The frequent item set mining method of the Apriori algorithm is replaced by the FP-Tree algorithm, and the operation efficiency of the algorithm is effectively improved. And completing data filtering by adding space-time characteristic constraints, and adding a promotion degree evaluation method to filter invalid data in the strong association rule again. The method and the device can quickly analyze the congestion tendency and provide information for city slow congestion.
The technical scheme adopted by the invention for solving the technical problems is as follows:
an algorithm improved congestion propagation analysis method comprises the following steps:
a) The congestion judging module processes data from the region, divides the traffic state by calculating the intersection saturation, and calculates the saturation of each lane of the intersection by calculating the traffic demand index TI and the intersection traffic capacity so as to realize the division of the traffic state;
b) The item set generation module integrates the results obtained from the congestion judgment module;
c) Generating an FP-Tree frequent item set;
d) Screening an item set by using an Apriori algorithm, excavating a congestion frequent item set, and analyzing the propagation rule of road congestion according to the congestion frequent item set;
e) The purpose of congestion tendency analysis is achieved through a congestion tendency analysis module.
Specifically, the traffic demand index TI is calculated by first obtaining road network topology data, and obtaining data corresponding to all fields according to each intersection included in a region; calculating the total traffic volume, occupancy and speed of the 5min traffic of the lane and the previous two time interval values through the lane number and the license plate information, and then calculating the traffic demand index TI; then calculating the traffic capacity of each lane of the intersection, and calculating according to the number of lanes at the intersection, channelized information and corresponding traffic flow data per minute; finally, calculating the saturation of each lane of the intersection according to the result, carrying out normalization processing on the saturation, wherein the value range is [0,1], and judging the traffic state of the intersection according to the level divided by the saturation;
the formula of the traffic demand index TI is,
in the formula, N represents the current time, k represents the kth lane of the current intersection, F represents the standardized traffic volume, and the time interval of calculating the TI value is 0.5 second; v represents a congestion determination speed, and is 5km/h as a default;the TI value is the kth lane at the current intersection and the time N;obtaining smooth traffic volume for the kth lane of the current intersection;is the smoothed occupancy;is the smoothing speed; alpha is alpha k The traffic weight coefficient of the kth lane of the current intersection; beta is a k The occupancy weight coefficient of the k-th lane of the current intersection; gamma ray k The speed weight coefficient of the kth lane of the current intersection, alpha is the traffic weight coefficient, beta is the occupancy weight coefficient, and gamma is the speed weight coefficient;
the formula I, the formula II and the formula III are respectively the smoothed traffic volume, the occupancy and the speed, wherein N is the current time; t is the time interval, T =5min; a is a weighting coefficient of a value before two time intervals; b is a weighting coefficient of a value before a time interval; c is a current value weight coefficient;
the lane traffic capacity refers to the maximum number of vehicles which can pass through the cross section in unit time detected by the detector, and the traffic capacity calculation formula of each lane at the intersection is a fourth formula;
and IV, formula IV: c i =max(q t );
In the formula IV, C i The traffic capacity of the ith lane is shown, and i is the lane; q. q.s t For the number of vehicles passing through the lane, t is a positive integer and belongs to [1,5 ]]In minutes;
the calculation formula of the saturation is formula five;
in the fifth formula, B is the saturation degree of the intersection; TI is a traffic demand index with a time interval of 5min; c is the traffic capacity calculated by taking 5min as a unit;
the formula for normalizing B is six
specifically, the item set generation module works in a manner that,
a, B, C represent three intersections, 1001, 1002 and 1003, respectively; 1 and 2 represent different lanes in the same intersection; let T denote time, T = { T = } 1 ,T 2 ,T 3 ,...,T i },T i The time interval of the ith moment and the two moments is 5 minutes;
for the set of transaction items at intersection a, the formula seven needs to be satisfied: a = { A = 1 ,A 2 ,A 3 ,...,A n };
In the formula VII, A 1 、A 2 、A 3 Respectively represent the traffic states of 1, 2 and 3 lanes at the intersection A, A n Representing the traffic state of the nth lane of the intersection A;
generating a multi-element time series combined set according to the transaction group generated according to the time series, wherein the combined set comprises the following components:
T 3 corresponds to { B 2 (3)}、T 4 Corresponds to { B 1 (3),B 2 (3),C 2 (3)}、T 5 Corresponds to { A 1 (3),B 1 (3),B 2 (3),C 2 (3)}、T 6 Corresponds to { A 1 (3),B 1 (3),B 2 (3),C 2 (4)}、T 7 Corresponds to { A 1 (3),A 2 (3),B 1 (4),B 2 (4),C 1 (3),C 2 (4)}、T 8 Corresponds to { A 1 (4),A 2 (3),B 1 (3),B 2 (4),C 1 (3),C 2 (4)};
Each item in the merged set of the multivariate time series represents the traffic state grade of a certain lane of a certain intersection;
grouping according to the item set time numbers of the transaction items, and sliding downwards at 1 interval every 3 times to generate the following time sliding window transaction sets:
T′ 1 corresponds to B 2 (3) 3 ;
T′ 2 Corresponds to B 2 (3) 2 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 3 Corresponds to B 2 (3) 1 、B 1 (3) 2 、B 2 (3) 2 、C 2 (3) 2 、A 1 (3) 3 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 4 Corresponds to B 1 (3) 1 、B 2 (3) 1 、C 2 (3) 1 、A 1 (3) 2 、B 1 (3) 2 、B 2 (3) 2 、C 2 (3) 2 、A 1 (3) 3 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 5 Corresponds to A 1 (3) 1 、B 1 (3) 1 、B 2 (3) 1 、C 2 (3) 1 、A 1 (3) 2 、B 1 (3) 2 、B 2 (3) 2 、C 2 (4) 2 、A 1 (3) 3 、A 2 (3) 3 、B 1 (4) 3 、B 2 (4) 3 、C 1 (3) 3 、C 2 (4) 3 ;
T′ 6 Corresponds to A 1 (3) 1 、B 1 (3) 1 、B 2 (3) 1 、C 2 (4) 1 、A 1 (3) 2 、A 2 (3) 2 、B 1 (4) 2 、B 2 (4) 2 、C 1 (3) 2 、C 2 (4) 2 、A 1 (4) 3 、A 2 (3) 3 、B 1 (3) 3 、B 2 (4) 3 、C 1 (3) 3 、C 2 (4) 3 ;
Each item in the time sliding window transaction set represents the traffic state grade of a certain lane at a certain intersection, and the 1 st, 2 nd and 3 rd moments after grouping according to the sliding window are marked; t ' represents a time number after the sliding window, and T ' includes T ' 1 ~T′ 6 Is a reaction of T 1 ~T 8 And generating a new time number after sliding according to the sliding window size of 3 and the interval of 1.
Specifically, the FP-Tree frequent item set is generated in a mode that,
step one, firstly establishing an item head, and establishing a FP tree item head as follows:
item A 1 (3) Has a frequency of 3, term A 1 (4) The frequency of (1) and the term A 2 (3) Has a frequency of 2, item B 1 (3) The frequency of (2) is 4, item B 1 (4) The frequency of (1) and the term B 2 (3) Has a frequency of 4, item B 2 (4) The frequency of (1) is 2, item C 1 (3) The frequency of (1) is 2, item C 2 (3) The frequency of (1) is 2, item C 2 (4) The frequency of (3);
secondly, scanning data to obtain the counts of all frequent 1 item sets; deleting the items with the promotion degree L less than or equal to 1, putting the 1 item frequent sets into an item head table, and arranging the items in a descending order according to the support degree;
thirdly, scanning data, removing the non-frequent item sets from the read original data, and arranging the non-frequent item sets in a descending order according to the support degree;
step four, reading in the sorted data set, inserting the FP tree, and inserting the FP tree into the FP tree according to the sorted sequence during insertion, wherein the node in the front of the order is an ancestor node, and the node in the back of the order is a descendant node; if there is a common ancestor, the count of the corresponding common ancestor node is increased by 1; after insertion, when a new node appears, the node corresponding to the item head can be linked with the new node through the node linked list until all data are inserted into the FP tree, and the establishment of the FP tree is completed;
step five, finding the conditional mode base corresponding to the item head from the bottom item of the item head upwards in sequence, and obtaining a frequent item set from the recursive mining of the conditional mode base;
and step six, if the number of the terms of the frequent term set is not limited, returning to the step five, wherein all the frequent term sets are returned, otherwise, only returning the frequent term set meeting the requirement of the number of the terms.
Specifically, apriori algorithm describes the effectiveness of the rule by using two index parameters, namely, support degree and confidence degree, and the support degree calculation formula is as follows:s in the support degree calculation formula represents the support degree; x, Y represents two items in an item set, and X @ Y represents the union of the two items; σ represents the number of tuples present in a set of transaction groups; n represents the total number of tuples of the transaction database;
the confidence coefficient calculation formula is as follows:c in the confidence coefficient calculation formula represents the confidence coefficient; the calculation of confidence depends on the support; wherein, C (X → Y) ≠ C (Y → X);
after the minimum confidence threshold filtering is designed, the invalid data in the strong association rule is filtered again by using the promotion degree, and the calculation formula of the promotion degree is as follows:in the lifting degree calculation formula, L represents the lifting degree; the threshold for the degree of lift is set to 1,L>1, indicating that the positive correlation is higher; l is<1, indicating that the higher the negative correlation; l =1, then no correlation is indicated; invalid association rules in strong association rules are effectively filtered by filtering tuples satisfying L ≦ 1.
Specifically, the congestion tendency analysis module comprises an electronic monitoring device, a processor and a memory, wherein the electronic monitoring device and the memory are electrically connected with the processor.
In particular, in step a), the data in the area comprises detector and electrical alarm data.
The beneficial effects of the invention are: the invention provides an algorithm improved congestion propagation analysis method. The frequent item set mining method of the Apriori algorithm is replaced by the FP-Tree algorithm, and the operation efficiency of the algorithm is effectively improved. And completing data filtering by adding space-time characteristic constraints, and adding a promotion degree evaluation method to filter invalid data in the strong association rule again. The method and the device can rapidly analyze the congestion tendency and provide information for city congestion relief.
Drawings
The invention is further described with reference to the following figures and examples.
FIG. 1 is a block diagram of the present invention;
FIG. 2 is a block diagram of the FP tree of the present invention;
FIG. 3 is a block diagram of a congestion tendency analysis module according to the present invention;
Detailed Description
The invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams each illustrating only the basic structure of the invention in a schematic manner, and therefore show only the constitution related to the invention.
FIG. 1 is a block diagram of the present invention; FIG. 2 is a block diagram of the FP tree of the present invention; fig. 3 is a schematic structural diagram of a congestion tendency analysis module according to the present invention.
The data preprocessing module mainly comprises a congestion judging module and an item set generating module, wherein the congestion judging module mainly judges the congestion state of intersections in a road network and calculates the saturation of each lane of the intersections according to intersection detector data. Calculating the saturation degree by depending on a Traffic demand Index (Traffic Index, hereinafter referred to as TI) and the Traffic capacity of lanes, calculating the saturation degree of each lane of the intersection, dividing the congestion degree of each lane of the intersection according to the saturation degree, and evaluating the congestion state of each lane of the intersection by four congestion degrees, namely smooth congestion, light congestion, medium congestion and heavy congestion; the item set generation module is mainly used for grading the congestion degree obtained by the congestion judgment module and adding the constraints of the road network topology and the intersection congestion occurrence time to assist in filtering the invalid item set. The congestion item set establishes a road network congestion item set of three items of smooth traffic, moderate congestion and severe congestion, wherein the mild congestion is divided into smooth levels.
The improved Apriori algorithm in the application is to replace the mining algorithm of a frequent item set with an FP-Tree algorithm. The frequent item set mining method of the Apriori algorithm needs to scan data for multiple times and has a large I/O port bottleneck, but the FP-Tree algorithm only needs to scan the data sets twice, so that the operation efficiency of the algorithm can be effectively improved; the association rule generation and filtering module is additionally provided with a promotion degree method on the basis of continuing a support degree and confidence degree method in an Apriori algorithm, the generated strong association rule data are subjected to fine screening, invalid association rule data in the strong association rule data are filtered, and the efficiency and accuracy of the algorithm are further improved.
The congestion tendency analysis module is mainly used for applying the congestion propagation rules excavated by the modules to the road network area so as to analyze the propagation of the congestion at the intersections and obtain the time sequence of the congestion at each intersection in the road network. When congestion occurs at a certain congestion source intersection in the road network, the subsequent congestion probability of other intersections in the rule is analyzed according to the congestion propagation trend in the congestion propagation rule, and therefore auxiliary decision information is provided for the city.
As shown in fig. 1, the congestion judging module processes data of detectors and electric alarms from an area, divides traffic states by calculating intersection saturation, and calculates each lane saturation of an intersection by calculating a traffic demand index TI and intersection traffic capacity, thereby realizing the division of the traffic states.
The calculation method of the traffic demand index TI comprises the steps of firstly obtaining road network topological data, and obtaining data corresponding to all fields according to each intersection contained in a region; calculating the total traffic volume, occupancy and speed of the 5min traffic of the lane and the previous two time interval values through the lane number and the license plate information, and then calculating the traffic demand index TI; generally, the traffic data of each lane at each intersection in a road network is uploaded every 1 minute by default for the electric police data, and the calculation granularity of the text TI is 5 minutes; then calculating the traffic capacity of each lane of the intersection, and calculating according to the number of lanes at the intersection, channelized information and corresponding traffic flow data per minute; finally, calculating the saturation of each lane of the intersection according to the result, carrying out normalization processing on the saturation, wherein the value range is [0,1], and judging the traffic state of the intersection according to the level divided by the saturation;
the formula of the traffic demand index TI is,
in the formula, N represents the current time, k represents the lane of the current intersection, F represents the standardized traffic volume, and the default is to calculate the TI value time interval (second)/2; v represents a congestion determination speed, and is 5km/h as a default;the TI value is the kth lane at the current intersection and the time N;obtaining smooth traffic volume for the kth lane of the current intersection;is the smoothed occupancy;is the smoothing speed; alpha is alpha k The weight coefficient of the traffic volume of the k lanes at the current intersection is obtained; beta is a k The occupancy weight coefficient of the k lanes at the current intersection is obtained; gamma ray k The speed weight coefficient of a K lane at the current intersection is shown, alpha is a traffic weight coefficient, beta is an occupancy weight coefficient, and gamma is a speed weight coefficient;
the formula I, the formula II and the formula III are respectively the smoothed traffic volume, the occupancy and the speed, wherein N is the current time; t is the time interval, T =5min; a is a weighting coefficient of a value before two time intervals; b is a weighting coefficient of a value before a time interval; c is a current value weight coefficient;
the lane traffic capacity refers to the maximum number of vehicles which can pass through the cross section in unit time detected by the detector, and the traffic capacity calculation formula of each lane at the intersection is a fourth formula;
and IV, formula IV: c i =max(q t );
In the formula IV, C i The traffic capacity of the ith lane is shown, and i is the lane; q. q.s t For the number of vehicles passing through the lane, t is a positive integer, and t belongs to [1,5 ]]In minutes;
the calculation formula of the saturation is formula five;
in the fifth formula, B is the saturation of the intersection; TI is a traffic demand index with a time interval of 5min; c is the traffic capacity calculated by taking 5min as a unit;
the formula for normalizing B is shown as six.
the traffic states are divided according to the normalized saturation in table 1.
Grade | Degree of saturation | Traffic state |
1 | B≤0.7 | Clear |
2 | 0.7<B≤0.8 | |
3 | 0.8<B≤0.9 | |
4 | 0.9<B≤1 | Severe congestion |
TABLE 1
Through the formula and the table 1, the saturation information of each lane can be obtained, and the traffic state information of the corresponding intersection can be obtained by performing state division according to the intersection traffic state division table in the table 1. In consideration of the division of four traffic states of clear, light, medium, and heavy congestion 1, 2, 3, and 4, respectively, 1001, 1002, and 1003 in the table indicate intersection numbers of intersections, respectively. Assuming that two cases of going straight and turning left at each intersection are considered, the output samples are shown in table 2.
TABLE 2
And the item set generation module integrates the results obtained from the congestion judgment module. When the item set is generated, in order to save memory and improve algorithm efficiency, the smooth and light congestion states are ignored, and only the moderate congestion and heavy congestion states of 3 and 4 grades are considered. A, B, C denote three intersections 1001, 1002, and 1003, respectively, and 1 and 2 denote different lanes in the same intersection.
Let T denote time, T = { T = } 1 ,T 2 ,T 3 ,...,T i },T i The time interval between the ith time and the two preceding and succeeding times is 5 minutes. The integration is shown in Table 3.
TABLE 3
When constructing the set of items from an actual intersection, all lanes of the entire intersection need to be considered. For example, for the set of transaction items at intersection a, it is required to satisfy a = { a = { (a) } 1 ,A 2 ,A 3 ,...,A n }。
In the formula, A 1 、A 2 And A 3 Respectively represent traffic states of 1, 2 and 3 lanes at intersection A, A n Traffic pattern showing the nth lane of intersection AState.
From the transaction groups generated in time series, a merged set of multivariate time series is generated, see table 4,
T | item set |
T 1 | —— |
T 2 | —— |
T 3 | {B 2 (3)} |
T 4 | {B 1 (3),B 2 (3),C 2 (3)} |
T 5 | {A 1 (3),B 1 (3),B 2 (3),C 2 (3)} |
T 6 | {A 1 (3),B 1 (3),B 2 (3),C 2 (4)} |
T 7 | {A 1 (3),A 2 (3),B 1 (4),B 2 (4),C 1 (3),C 2 (4)} |
T 8 | {A 1 (4),A 2 (3),B 1 (3),B 2 (4),C 1 (3),C 2 (4)} |
TABLE 4
Each entry in the set of entries in table 4 represents the traffic status level for a lane at an intersection.
Considering the congestion propagation law, namely: congestion at a current intersection is likely to cause congestion at adjacent intersections after a period of time. Therefore, grouping is performed according to the item set time numbers of the transaction items, and downward sliding is performed at 1 interval every 3 times, so as to generate the following time sliding window transaction sets, which is shown in table 5:
TABLE 5
Each entry in the entry set of table 5 represents: a traffic status level of a lane at an intersection that appears in a time sliding window sequence; the superscripts are the 1 st, 2 nd, 3 rd time after grouping by the sliding window.
Such as: b 2 (3) 1 The congestion level of the 2 nd lane of the intersection B at the 1 st moment in the sliding window is represented as 3 levels; c 1 (3) 2 The congestion level of the 1 st lane of the C intersection at the 2 nd time point at the representative sliding window is 3 levels.
The size of the sliding window is 3, so that T in the multi-element time sequence merging set can be filtered 1 And T 2 When the time item set is empty, the filtering is incomplete if the sliding number is less than 3, and there are still empty item sets, and if the sliding number is too large, the item sets are too complicated to analyze, so the sliding number is set to 3 and the sliding interval is 1. T ' in Table 5 represents a time number after the sliding window, and T ' includes T ' 1 ~T′ 6 Is a reaction of T 1 ~T 8 And generating a new time number after sliding according to the sliding window size of 3 and the interval of 1.
(2) The FP-Tree frequent item set is generated, a large number of item and item combinations are generally generated according to an Apriori algorithm frequent item set mining algorithm, and the frequency of the sub item sets formed is realized through multiple circulation traversal. The device considers that the FP-Tree algorithm is used for replacing an Apriori algorithm frequent item set mining algorithm, the FP-Tree algorithm only needs to scan data sets twice, and the operation efficiency of the algorithm can be effectively improved.
To construct a frequent item set using the FP-Tree algorithm, a term header is first established. Sample item header table is shown in table 6:
item(s) | Frequency of |
A 1 (3) | 3 |
A 1 (4) | 1 |
A 2 (3) | 2 |
B 1 (3) | 4 |
B 1 (4) | 1 |
B 2 (3) | 4 |
B 2 (4) | 2 |
C 1 (3) | 2 |
C 2 (3) | 2 |
C 2 (4) | 3 |
TABLE 6
Sort by frequency according to the item head table, see table 7:
as shown in fig. 2, a FP-tree is built according to the sorted item head table and item set.
After the FP-Tree is built, when a new node is inserted, the node corresponding to the item head table is connected with the new node through the node chain table until the insertion of all data is completed, and the building of the FP Tree is completed. After the FP tree is built, the conditional mode bases of all leaf nodes are built one by one according to C 2 (4) For example, C 2 (4) See table 8 for the conditional mode bases:
item(s) | Frequency of |
A 2 (3) | 2 |
B 2 (4) | 2 |
C 1 (3) | 2 |
TABLE 8
In this way, the two sets of terms are recursively combined to obtain the final frequent set of terms.
Based on the above analysis, such as a two-item frequent item set: { B 1 (3):3,B 2 (3):3},...;
Three frequent item sets: { A 1 (3):2,B 1 (3):2,B 2 (3):2},...;
Four frequent itemsets: { A) 2 (3):2,B 2 (4):2,C 1 (3):2,C 2 (4):2}。
The frequent item sets of two items and three items are not complete, and the processing is omitted. The most frequent item set finally analyzed in this example is a frequent four item set, that is: { A) 2 (3):2,B 2 (4):2,C 1 (3):2,C 2 (4):2}。
Step one, firstly establishing an item head, and establishing a FP tree item head as follows:
item A 1 (3) Has a frequency of 3, term A 1 (4) The frequency of (1) and the term A 2 (3) Has a frequency of 2, item B 1 (3) Has a frequency of 4, item B 1 (4) The frequency of (1) and the term B 2 (3) Has a frequency of 4, item B 2 (4) The frequency of (1) is 2, item C 1 (3) The frequency of (2) and the term C 2 (3) The frequency of (1) is 2, item C 2 (4) The frequency of (3);
secondly, scanning data to obtain the counts of all frequent 1 item sets; deleting items with the lifting degree L less than or equal to 1, putting the 1 item frequent sets into an item head table, and arranging the items in a descending order according to the support degree;
thirdly, scanning data, removing the non-frequent item sets from the read original data, and arranging the items in a descending order according to the support degree;
step four, reading in the sorted data set, inserting the FP tree, and inserting the FP tree into the FP tree according to the sorted sequence during insertion, wherein the node in the front of the order is an ancestor node, and the node in the back of the order is a descendant node; if there is a common ancestor, the count of the corresponding common ancestor node is increased by 1; after insertion, when a new node appears, the node corresponding to the item head can be linked with the new node through the node linked list until all data are inserted into the FP tree, and the establishment of the FP tree is completed;
step five, finding the condition mode base corresponding to the item head from the bottom item of the item head upwards in sequence, and recursively mining the condition mode base to obtain a frequent item set;
and step six, if the number of the terms of the frequent term set is not limited, returning to the step five, wherein all the frequent term sets are returned, otherwise, only returning the frequent term set meeting the requirement of the number of the terms.
The Apriori algorithm describes the effectiveness of the rule by two index parameters of support degree and confidence degree, namely, the Apriori algorithm is used for screening an item set, mining a congestion frequent item set and analyzing the propagation rule of road congestion according to the congestion frequent item set. The support calculation formula is as follows:the rules may be determined for how often a given data set is to be used. S in the support degree calculation formula represents the support degree; x, Y represents two items in an item set, and X @ Y represents the union of the two items; σ represents the number of tuples present in a set of transaction groups; n represents the total number of tuples of the transaction database;
the confidence coefficient calculation formula is as follows:the frequency with which Y occurs in transactions containing X is determined. C in the confidence coefficient calculation formula represents a confidence coefficient; the calculation of confidence depends on the support; wherein, C (X → Y) ≠ C (Y → X);
after the minimum confidence threshold filtering is designed, the invalid data in the strong association rule is filtered again by using the promotion degree, and the calculation formula of the promotion degree is as follows:y is the degree of promotion in the transaction group containing X. In the lifting degree calculation formula, L represents the lifting degree; the threshold for the degree of lift is set to 1,L>1, indicating that the positive correlation is higher; l is<1, indicating that the negative correlation is higher; l =1, then no correlation is indicated; invalid association rules in strong association rules are effectively filtered by filtering tuples satisfying L ≦ 1.
Road network topological constraint:
the road network topology constraint in the scheme is mainly used for filtering irrelevant data such as a rule that intersection distances are far in space and congestion occurs at the same time, limiting the spatial range of intersection congestion propagation at adjacent intersections, considering that each index of the intersection is calculated every 5 minutes, and limiting the time range at congestion propagation time intervals within 15 minutes.
The FP trees are more simplified by matching the promotion degree and the road network topological constraint with the established FP trees, the pruning effect is achieved, and a large amount of storage space is saved. And the characteristic of high retrieval efficiency of the FP tree is combined, so that the operation efficiency of the algorithm is greatly improved.
Timing constraints. The timing sequence constraint condition in the scheme mainly sequences the time of intersection congestion occurrence, and filters data according to the sequence of the congestion occurrence time. Such as: if the south-north straight-going direction of the intersection A is heavily congested at the time of 6:45 of the day, the south-north straight-going direction of the intersection B in the south-north direction of the intersection A is congested in a rule analysis that the congestion is not taken as a congestion propagation phenomenon in the day 16. I.e., timing constraints are primarily intended to filter out invalid rules that violate a rule.
(3) And (5) analyzing the congestion propagation result. The improved Apriori algorithm provided by the invention is used for mining the congestion propagation rule of the intersection, and the congestion propagation rule is obtained by taking the early peak time period of a certain region of Harbin as an example. The congestion propagation law mining is shown in table 10:
watch 10
And obtaining a congestion propagation rule by analyzing the space-time characteristics.
As shown in fig. 3, the congestion tendency analysis module excavates a congestion propagation rule of the intersections in the urban area according to the above contents, and can obtain the influence of congestion diffusivity of all monitored intersections in the urban area. And (4) obtaining a congestion propagation rule of two dimensions of time sequence and space of the urban area through analysis. When the congestion phenomenon occurs in a certain intersection, the probability that the congestion phenomenon of different degrees will occur to other intersections connected with the intersection after a period of time in the same direction is given, so that the congestion propagation occurrence trend of the intersections in the urban area is obtained, the function of analyzing the congestion trend is realized, the reference is given to relevant departments, and the decision of delaying congestion is assisted.
Claims (3)
1. An algorithm improved congestion propagation analysis method is characterized by comprising the following steps: the method comprises the following steps:
a) The congestion judging module processes data from the region, divides the traffic state by calculating the intersection saturation, and calculates the saturation of each lane of the intersection by calculating the traffic demand index TI and the intersection traffic capacity so as to realize the division of the traffic state;
b) The item set generation module integrates the results obtained from the congestion judgment module;
c) Generating an FP-Tree frequent item set;
d) Screening an item set by using an Apriori algorithm, excavating a congestion frequent item set, and analyzing the propagation rule of road congestion according to the congestion frequent item set;
e) The purpose of congestion tendency analysis is achieved through a congestion tendency analysis module;
the calculation method of the traffic demand index TI comprises the steps of firstly obtaining road network topological data, and obtaining data corresponding to all fields according to each intersection contained in a region; calculating the total traffic volume, occupancy and speed of the 5min traffic of the lane and the previous two time interval values through the lane number and the license plate information, and then calculating the traffic demand index TI; then calculating the traffic capacity of each lane of the intersection, and calculating according to the number of lanes at the intersection, channelized information and corresponding traffic flow data per minute; finally, calculating the saturation of each lane of the intersection according to the result, carrying out normalization processing on the saturation, wherein the value range is 0,1, and judging the traffic state of the intersection according to the grade divided by the saturation;
the formula of the traffic demand index TI is,
in the formula, N represents the current time, k represents the kth lane of the current intersection, F represents the standardized traffic volume, and the time interval of calculating the TI value is 0.5 second; v represents a congestion determination speed, and is 5km/h as a default;the TI value is the kth lane at the current intersection and the time N;obtaining smooth traffic volume for the kth lane of the current intersection;is the smoothed occupancy;is the smoothing speed; alpha is alpha k Is as followsThe traffic weight coefficient of the kth lane of the front intersection; beta is a beta k The occupancy weight coefficient of the k-th lane of the current intersection; gamma ray k The speed weight coefficient of the kth lane of the current intersection, alpha is the traffic weight coefficient, beta is the occupancy weight coefficient, and gamma is the speed weight coefficient;
the formula I, the formula II and the formula III are respectively the traffic volume, the occupancy and the speed after smoothing, wherein N is the current time; t is the time interval, T =5min; a is a weighting coefficient of a value before two time intervals; b is a weighting coefficient of a value before a time interval; c is a current value weight coefficient;
the lane traffic capacity refers to the maximum number of vehicles which can pass through the cross section in unit time detected by the detector, and the traffic capacity calculation formula of each lane at the intersection is a fourth formula;
the formula IV is as follows: c i =max(q t );
In the formula IV, C i The traffic capacity of the ith lane is, and i is the lane; q. q.s t For the number of vehicles passing through the lane, t is a positive integer and belongs to [1,5 ]]In minutes;
the calculation formula of the saturation is formula five;
in the fifth formula, B is the saturation of the intersection; TI is a traffic demand index with a time interval of 5min; c is the traffic capacity calculated by taking 5min as a unit;
the formula for normalizing B is six
the manner in which the item set generation module operates is,
a, B, C are representative of three intersections, 1001, 1002 and 1003, respectively; 1 and 2 represent different lanes in the same intersection; let T denote the time, T = { T = 1 ,T 2 ,T 3 ,...,T i },T i The time interval of the ith moment and the two moments is 5 minutes;
for the set of transaction items at intersection a, the formula seven needs to be satisfied: a = { A = 1 ,A 2 ,A 3 ,...,A n };
In the formula VII, A 1 、A 2 、A 3 Respectively represent the traffic states of 1, 2 and 3 lanes at the intersection A, A n Representing the traffic state of the nth lane of the intersection A;
generating a multi-element time sequence merging set according to the transaction group generated according to the time sequence, wherein the multi-element time sequence merging set comprises the following steps:
T 3 corresponds to { B 2 (3)}、T 4 Corresponds to { B 1 (3),B 2 (3),C 2 (3)}、T 5 Corresponds to { A 1 (3),B 1 (3),B 2 (3),C 2 (3)}、T 6 Corresponds to { A 1 (3),B 1 (3),B 2 (3),C 2 (4)}、T 7 Corresponds to { A 1 (3),A 2 (3),B 1 (4),B 2 (4),C 1 (3),C 2 (4)}、T 8 Corresponds to { A 1 (4),A 2 (3),B 1 (3),B 2 (4),C 1 (3),C 2 (4)};
Each item in the combined set of the multivariate time series represents the traffic state grade of a certain lane of a certain intersection;
grouping according to the item set time numbers of the transaction items, and sliding downwards at 1 interval every 3 times to generate the following time sliding window transaction sets:
T′ 1 corresponds to B 2 (3) 3 ;
T′ 2 Corresponds to B 2 (3) 2 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 3 Corresponds to B 2 (3) 1 、B 1 (3) 2 、B 2 (3) 2 、C 2 (3) 2 、A 1 (3) 3 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 4 Corresponds to B 1 (3) 1 、B 2 (3) 1 、C 2 (3) 1 、A 1 (3) 2 、B 1 (3) 2 、B 2 (3) 2 、C 2 (3) 2 、A 1 (3) 3 、B 1 (3) 3 、B 2 (3) 3 、C 2 (3) 3 ;
T′ 5 Corresponds to A 1 (3) 1 、B 1 (3) 1 、B 2 (3) 1 、C 2 (3) 1 、A 1 (3) 2 、B 1 (3) 2 、B 2 (3) 2 、C 2 (4) 2 、A 1 (3) 3 、A 2 (3) 3 、B 1 (4) 3 、B 2 (4) 3 、C 1 (3) 3 、C 2 (4) 3 ;
T′ 6 Corresponds to A 1 (3) 1 、B 1 (3) 1 、B 2 (3) 1 、C 2 (4) 1 、A 1 (3) 2 、A 2 (3) 2 、B 1 (4) 2 、B 2 (4) 2 、
C 1 (3) 2 、C 2 (4) 2 、A 1 (4) 3 、A 2 (3) 3 、B 1 (3) 3 、B 2 (4) 3 、C 1 (3) 3 、C 2 (4) 3 ;
Each item in the time sliding window transaction set represents the traffic state grade of a certain lane at a certain intersection, and the 1 st, 2 nd and 3 rd moments after grouping according to the sliding window are marked; t ' represents a time number after the sliding window, and T ' includes T ' 1 ~T′ 6 Is a reaction of T 1 ~T 8 Generating a new time number after sliding according to the sliding window size of 3 and the interval of 1;
the FP-Tree frequent item set is generated in a mode that,
firstly, establishing an item head, wherein the item head of the FP tree is established as follows:
item A 1 (3) Has a frequency of 3, term A 1 (4) The frequency of (A) is 1, item A 2 (3) Has a frequency of 2, item B 1 (3) Has a frequency of 4, item B 1 (4) The frequency of (1) and the term B 2 (3) Has a frequency of 4, item B 2 (4) The frequency of (2) and the term C 1 (3) The frequency of (1) is 2, item C 2 (3) The frequency of (1) is 2, item C 2 (4) The frequency of (2) is 3;
secondly, scanning data to obtain the counts of all frequent 1 item sets; deleting the items with the promotion degree L less than or equal to 1, putting the 1 item frequent sets into an item head table, and arranging the items in a descending order according to the support degree;
thirdly, scanning data, removing the non-frequent item sets from the read original data, and arranging the items in a descending order according to the support degree;
step four, reading in the sorted data set, inserting a FP tree, and inserting the FP tree into the FP tree according to the sorted order during insertion, wherein the node in the front of the sorting is an ancestor node, and the node in the back of the sorting is a descendant node; if there is a common ancestor, the count of the corresponding common ancestor node is increased by 1; after insertion, when a new node appears, the node corresponding to the item head can be linked with the new node through the node linked list until all data are inserted into the FP tree, and the establishment of the FP tree is completed;
step five, finding the condition mode base corresponding to the item head from the bottom item of the item head upwards in sequence, and recursively mining the condition mode base to obtain a frequent item set;
step six, if the number of terms of the frequent item set is not limited, returning to the step five, wherein all the frequent item sets are returned, otherwise, only returning the frequent item set meeting the requirement of the number of terms;
the Apriori algorithm describes the effectiveness of the rule by two index parameters of support degree and confidence degree, and the calculation formula of the support degree is as follows:s in the support degree calculation formula represents the support degree; x, Y represents two items in an item set, and X @ Y represents the union of the two items; σ represents the number of tuples present in a set of transaction groups; n represents the total number of tuples of the transaction database;
the confidence coefficient calculation formula is as follows:c in the confidence coefficient calculation formula represents a confidence coefficient; the calculation of confidence depends on the support; wherein, C (X → Y) ≠ C (Y → X);
after the minimum confidence coefficient threshold filtering is designed, the invalid data in the strong association rule is filtered again by using the promotion degree, and the calculation formula of the promotion degree is as follows:in the lifting degree calculation formula, L represents the lifting degree; the threshold for the degree of lift is set to 1,L>1, indicating that the positive correlation is higher; l is<1, indicating that the higher the negative correlation; l =1, then no correlation is indicated; invalid association rules in strong association rules are effectively filtered by filtering tuples satisfying L ≦ 1.
2. The algorithm-improved congestion propagation analysis method according to claim 1, characterized in that: the congestion tendency analysis module comprises electronic monitoring equipment, a processor and a memory, wherein the electronic monitoring equipment and the memory are electrically connected with the processor.
3. The method for analyzing congestion propagation based on algorithm improvement as claimed in claim 1, wherein: in step a), the data in the area includes detector and electrical alarm data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111381097.1A CN114170796B (en) | 2021-11-20 | 2021-11-20 | Algorithm improved congestion propagation analysis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111381097.1A CN114170796B (en) | 2021-11-20 | 2021-11-20 | Algorithm improved congestion propagation analysis method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114170796A CN114170796A (en) | 2022-03-11 |
CN114170796B true CN114170796B (en) | 2023-01-10 |
Family
ID=80479717
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111381097.1A Active CN114170796B (en) | 2021-11-20 | 2021-11-20 | Algorithm improved congestion propagation analysis method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114170796B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117114116A (en) * | 2023-08-04 | 2023-11-24 | 北京杰成合力科技有限公司 | Root cause analysis method, medium and equipment based on machine learning |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4370718A (en) * | 1979-02-06 | 1983-01-25 | Chasek Norman E | Responsive traffic light control system and method based on conservation of aggregate momentum |
CN101739818A (en) * | 2009-11-19 | 2010-06-16 | 北京世纪高通科技有限公司 | Method and device for data mining of road traffic accident based on association rule |
JP2013045233A (en) * | 2011-08-23 | 2013-03-04 | Toyota Infotechnology Center Co Ltd | Platooning control apparatus and platooning control method |
CN108320506A (en) * | 2018-02-05 | 2018-07-24 | 青岛大学 | A kind of discovery method of the congestion period based on composite network |
CN108960498A (en) * | 2018-06-27 | 2018-12-07 | 沈阳理工大学 | A kind of traffic congestion prediction technique based on temporal sequence association rule |
CN110111576A (en) * | 2019-05-16 | 2019-08-09 | 北京航空航天大学 | A kind of urban transportation elastic index and its implementation based on space-time congestion group |
CN110288824A (en) * | 2019-05-20 | 2019-09-27 | 浙江工业大学 | Based on Granger causality road network morning evening peak congestion and mechanism of transmission analysis method |
CN110992698A (en) * | 2019-12-26 | 2020-04-10 | 北京工业大学 | Method for calculating association degree between intersections based on Apriori support degree and driving distance in weighting manner |
CN111008223A (en) * | 2019-10-21 | 2020-04-14 | 北京交通大学 | Regional traffic jam correlation calculation method based on space-time association rule |
CN111159247A (en) * | 2019-11-20 | 2020-05-15 | 北京交通大学 | Regional traffic jam propagation track mining method |
CN112015837A (en) * | 2020-08-24 | 2020-12-01 | 安徽科力信息产业有限责任公司 | Urban road high-frequency path analysis method, system and storage medium |
CN113268528A (en) * | 2021-06-01 | 2021-08-17 | 西北工业大学 | Multi-probability threshold frequent item set mining method and device for sensing data |
CN113535819A (en) * | 2021-07-15 | 2021-10-22 | 广州交信投科技股份有限公司 | Traffic situation perception analysis method and device, computer storage medium and terminal |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107562865A (en) * | 2017-08-30 | 2018-01-09 | 哈尔滨工业大学深圳研究生院 | Multivariate time series association rule mining method based on Eclat |
-
2021
- 2021-11-20 CN CN202111381097.1A patent/CN114170796B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4370718A (en) * | 1979-02-06 | 1983-01-25 | Chasek Norman E | Responsive traffic light control system and method based on conservation of aggregate momentum |
CN101739818A (en) * | 2009-11-19 | 2010-06-16 | 北京世纪高通科技有限公司 | Method and device for data mining of road traffic accident based on association rule |
JP2013045233A (en) * | 2011-08-23 | 2013-03-04 | Toyota Infotechnology Center Co Ltd | Platooning control apparatus and platooning control method |
CN108320506A (en) * | 2018-02-05 | 2018-07-24 | 青岛大学 | A kind of discovery method of the congestion period based on composite network |
CN108960498A (en) * | 2018-06-27 | 2018-12-07 | 沈阳理工大学 | A kind of traffic congestion prediction technique based on temporal sequence association rule |
CN110111576A (en) * | 2019-05-16 | 2019-08-09 | 北京航空航天大学 | A kind of urban transportation elastic index and its implementation based on space-time congestion group |
CN110288824A (en) * | 2019-05-20 | 2019-09-27 | 浙江工业大学 | Based on Granger causality road network morning evening peak congestion and mechanism of transmission analysis method |
CN111008223A (en) * | 2019-10-21 | 2020-04-14 | 北京交通大学 | Regional traffic jam correlation calculation method based on space-time association rule |
CN111159247A (en) * | 2019-11-20 | 2020-05-15 | 北京交通大学 | Regional traffic jam propagation track mining method |
CN110992698A (en) * | 2019-12-26 | 2020-04-10 | 北京工业大学 | Method for calculating association degree between intersections based on Apriori support degree and driving distance in weighting manner |
CN112015837A (en) * | 2020-08-24 | 2020-12-01 | 安徽科力信息产业有限责任公司 | Urban road high-frequency path analysis method, system and storage medium |
CN113268528A (en) * | 2021-06-01 | 2021-08-17 | 西北工业大学 | Multi-probability threshold frequent item set mining method and device for sensing data |
CN113535819A (en) * | 2021-07-15 | 2021-10-22 | 广州交信投科技股份有限公司 | Traffic situation perception analysis method and device, computer storage medium and terminal |
Non-Patent Citations (4)
Title |
---|
A Spatiotemporal Apriori Approach to Capture Dynamic Associations of Regional Traffic Congestion;DONG-FAN XIE;《IEEE Acess》;20191227;第8卷;全文 * |
基于FP_Growth 算法的路段拥堵分析;刘良旭;《宁波工程学院学报》;20160930;第28卷(第3期);全文 * |
基于轨迹数据的城市交通拥塞传播模式挖掘;陈振华;《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》;20191215(第12期);全文 * |
时空关联规则挖掘算法及其在ITS中的应用;夏英;《计算机科学》;20110930;第38卷(第9期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN114170796A (en) | 2022-03-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107610469B (en) | Day-dimension area traffic index prediction method considering multi-factor influence | |
CN109410577B (en) | Self-adaptive traffic control subarea division method based on space data mining | |
CN102629305B (en) | Feature selection method facing to SNP (Single Nucleotide Polymorphism) data | |
CN104346379A (en) | Method for identifying data elements on basis of logic and statistic technologies | |
CN113516228B (en) | Network anomaly detection method based on deep neural network | |
CN114170796B (en) | Algorithm improved congestion propagation analysis method | |
CN112149922A (en) | Method for predicting severity of accident in exit and entrance area of down-link of highway tunnel | |
CN109635010A (en) | A kind of user characteristics and characterization factor extract, querying method and system | |
CN112767684A (en) | Highway traffic jam detection method based on charging data | |
CN106570076A (en) | Computer text classification system | |
CN113808392B (en) | Method for optimizing traffic accident data under multi-source data structure | |
CN109389172B (en) | Radio signal data clustering method based on non-parameter grid | |
CN112561175A (en) | Traffic accident influence factor prediction method, device, equipment and storage medium | |
CN109101568A (en) | Traffic high-risk personnel recognition methods based on XgBoost algorithm | |
CN116090819A (en) | Power distribution network risk situation prediction method based on association rule | |
CN114189364B (en) | Network node path reduction and prediction method based on Markov chain | |
CN113516850B (en) | Pipeline traffic flow data acquisition method based on space syntactic analysis | |
CN114969143A (en) | Inter-city atmospheric pollutant space-time correlation characteristic analysis method based on data mining | |
CN114595628A (en) | Differential transformation method for volume fracturing of horizontal well | |
CN113934772A (en) | Adaptive histogram publishing method facing data stream sliding window | |
CN113642162A (en) | Simulation comprehensive analysis method for urban road traffic emergency plan | |
CN113535819A (en) | Traffic situation perception analysis method and device, computer storage medium and terminal | |
CN112183597A (en) | Unbalanced traffic accident severity analysis method based on spectral clustering | |
CN111951546A (en) | Method for quantifying safety influence range of congestion charging policy | |
Tao et al. | Detection and optimization of traffic networks based on Voronoi diagram |
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 |