CN104750800A - Motor vehicle clustering method based on travel time characteristic - Google Patents
Motor vehicle clustering method based on travel time characteristic Download PDFInfo
- Publication number
- CN104750800A CN104750800A CN201510129468.5A CN201510129468A CN104750800A CN 104750800 A CN104750800 A CN 104750800A CN 201510129468 A CN201510129468 A CN 201510129468A CN 104750800 A CN104750800 A CN 104750800A
- Authority
- CN
- China
- Prior art keywords
- cluster
- vehicle
- clustering
- vehicles
- travel time
- 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.)
- Pending
Links
Abstract
The invention relates to a motor vehicle clustering method based on the travel time characteristic. The motor vehicle clustering method based on the travel time characteristic comprises the steps that customs pass traffic information of vehicles is extracted from an intelligent distributed traffic customs pass database; the number of times of passing through a customs pass of each vehicle in each hour every day is counted, so that the time characteristic vectors of the vehicles are obtained; the number k of optimum clustering centers is determined according to the max-min distance criterion; the KMeans clustering algorithm is compiled according to the MapReduce algorithm, the number k of optimum clustering centers is substituted into the KMeans clustering algorithm, and clustering analysis is conducted on the time characteristic vectors of the vehicles by means of the KMeans clustering algorithm, so that a clustering result is obtained; vehicle behaviors are analyzed according to the clustering result. According to the motor vehicle clustering method based on the travel time characteristic, the behavior characteristics of the motor vehicles are effectively learnt by grouping the vehicles, the vehicles are clustered according to the travel rule, so that vehicle information is obtained rapidly, and a reference basis is provided for evaluating the volume of traffic of the motor vehicles in the local area scientifically and predicting the road traffic passing condition in the future; meanwhile, the reference basis is also provided for checking suspicious vehicles for the public security and traffic police department.
Description
Technical field
The present invention relates to urban transportation gate management technical field, especially a kind of motor vehicle clustering method based on travel time feature.
Background technology
Along with the development of Computer Storage and sensor technology, traffic intelligent bayonet socket is widely used, and plays an important role to urban traffic control.The data that bayonet socket gathers grow with each passing day, and how to utilize the large data resource of bayonet socket to become the hot issue of research fully.
The application of traditional bayonet socket is mainly used in public security and deploys to ensure effective monitoring and control of illegal activities, alarm violating the regulations etc., can realize violating the regulations candid photograph, and automatically identify vehicles peccancy, increase work efficiency, avoid the impact because traffic police's on-site law-enforcing may cause traffic.In the large market demand of bayonet socket at present, also not open bayonet socket application of classifying based on the motor vehicle of travel time feature, cannot obtain information of vehicles rapidly, data-handling efficiency is lower.
Summary of the invention
The object of the present invention is to provide a kind of temporal characteristics according to vehicle driving to vehicle classification, more scientifically for traffic programme provides reference; Substantially increase the efficiency of data processing, the motor vehicle clustering method based on travel time feature of possibility is created in the widespread use for the large data of bayonet socket.
For achieving the above object, present invention employs following technical scheme: a kind of motor vehicle clustering method based on travel time feature, the method comprises the step of following order:
(1) from the distributed data base of intelligent traffic bayonet socket, extract the bayonet socket traffic information of vehicle;
(2) add up the number of times crossing bayonet socket of each little period of each car in one day, obtain the temporal characteristics vector of vehicle;
(3) minimax distance criterion determination Optimal cluster center number k is adopted;
(4) write KMeans clustering algorithm by MapReduce algorithm, Optimal cluster center number k is substituted in KMeans clustering algorithm, use the temporal characteristics vector of KMeans clustering algorithm to vehicle to carry out cluster analysis, obtain cluster result;
(5) vehicle behavioural analysis is carried out according to cluster result.
Utilize the data in the distributed data base of Apache Pheonix middleware extraction intelligent traffic bayonet socket, carry out structuring to data, the data later to structuring set up index.
Cross bayonet socket number of times according to MapReduce algorithm calculating vehicle each little period in one day, obtain the temporal characteristics vector of vehicle, i.e. 24 dimensional vector V
t.
Described minimax distance criterion comprises the following steps:
(1) Schilling cluster centre number is 3, cluster is carried out to the temporal characteristics vector of vehicle, obtain 3 groups of clusters, average to often organizing cluster, this mean value is the cluster centre of this group cluster, distance between the cluster centre calculating three groups of clusters, and MMD value when getting that wherein minor increment is 3 as cluster centre number; In like manner, try to achieve that cluster centre number is 4, MMD value corresponding to 5,6,7,8,9,10;
(2) getting the maximum cluster centre number of MMD value is Optimal cluster center number k.
Distance between described cluster centre is the Euclidean distance in 24 dimension spaces between 2.
As shown from the above technical solution, the present invention from the magnanimity that traffic intelligent bayonet socket is collected cross car information extract this method pay close attention to license plate number, bayonet socket numbering, by information such as bayonet socket times, utilize the ability of MapReduce technology and the powerful Distributed Calculation of cluster, calculate the temporal characteristics vector of all vehicles, and cluster is carried out to the vehicle sample of magnanimity, by the trip characteristics analysis to vehicle, the vehicle cluster of similar features will be had.Simultaneously, this method has validity and accuracy, by the division to vehicle, effectively can understand the behavioural characteristic of motor vehicle, carrying out cluster according to trip rule to vehicle can quick obtaining information of vehicles, for this area automobile traffic amount carries out the assessment of science, and provide reference frame to the prediction of future trajectory traffic passage situation, simultaneously also for public security deparment and traffic police department's investigation suspected vehicles provides reference frame.
Accompanying drawing explanation
Fig. 1 is the serial computing process flow diagram of KMeans clustering algorithm.
Fig. 2 is MapReduce operating mechanism schematic diagram.
Fig. 3 is the schematic diagram utilizing minimax distance criterion to detect Optimal cluster center number.
Fig. 4 is the schematic diagram of KMeans clustering algorithm MapReduce parallelization.
Fig. 5 is Kmeans cluster result cluster centre point schematic diagram.
Embodiment
As shown in Figure 1, KMeans clustering algorithm is a kind of indirect clustering method based on similarity measurement between sample, and its algorithm steps is: from n data object, first select arbitrarily k object as initial cluster center; For other remaining objects, according to similarity---the distance of they and these cluster centres, respectively they are distributed to the most similar to it---representated by cluster centre---cluster; And then calculate the cluster centre of each obtained new cluster---the average of all objects in this cluster; Constantly repeat this process till canonical measure function starts convergence.
As shown in Figure 2, MapReduce algorithm steps is: large data sets is decomposed into hundreds of small data set splist, each or several data sets are respectively by the node of 1 in cluster, it is exactly generally the logical computing machine of a Daepori, executed in parallel Map calculation task, specify mapping ruler and generate intermediate result, then these intermediate results perform Reduce calculation task by a large amount of nodal parallel again---and specify reductive rule, form net result.At data input phase, JobTracker obtains the storage element information of data slice to be calculated on NameNode; In the Map stage, JobTracker assigns multiple TaskTracker complete Map processor active task and generate intermediate result; The mixing that the Shuffle stage completes results of intermediate calculations exchanges; JobTracker assigns TaskTracker to complete Reduce task; Notify after Reduce task completes that JobTracker and NameNode is to produce last Output rusults.
As shown in Figure 4, KMeans clustering algorithm carries out the method for MapReduce: start 1 time corresponding MapReduce computation process to 1 iteration every in serial algorithm, complete data and be recorded to the distance calculating of cluster centre and the calculating of new cluster centre, Fig. 4 describes KMeans clustering algorithm MapReduce parallelization implementation method, in order to applicable MapReduce computation module process, pending data record must be stored with row form, make pending data can burst by row, and data non-correlation between sheet, the environment that Slicing procedure is run by Map-Reduce completes, do not need to write code.Before Reduce task starts, can divide into groups for index with key value to the intermediate result of Map tasks carrying node this locality and sort, to improve the execution efficiency of Reduce task.
The design of a.Map function
The task of Map function has been that each calculating being recorded to central point distance also marks its new cluster classification belonged to again, it is input as the cluster centre of all record data to be clustered and last round of iteration (or initial clustering), input data record (key, value) right form is < line number, record row >; Each Map function reads in cluster centre description document, and each measuring point of Map function to input calculates the class center nearest apart from it, and does the mark of new classification; The form exporting intermediate result (key, value) right is < cluster category IDs, record attribute vector >.
The design of b.Reduce function
The task of Reduce function is that the intermediate result obtained according to Map function calculates the cluster centre made new advances, for next round Map-ReduceJob. input data (key, value) right form is < cluster category IDs, { record attribute vector set } >; A Reduce task given in the record (namely having the record of identical category ID) that all key are identical---the some number that cumulative key is identical and each record component and, ask the average of each component, obtain new cluster centre description document; Output rusults (key, value) right form is < cluster category IDs, mean vector > judges whether this cluster restrains: the cluster centre that more last round of Map-ReduceJob obtains and epicycle MapReduce Job cluster centre distance, if change is less than given threshold value, then algorithm terminates; Otherwise, then replace last round of hub file with the cluster centre file of epicycle, and start the MapReduce Job of a new round.
Based on a motor vehicle clustering method for travel time feature, comprising: (1) extracts the bayonet socket traffic information of vehicle from the distributed data base of intelligent traffic bayonet socket; (2) add up the number of times crossing bayonet socket of each little period of each car in one day, obtain the temporal characteristics vector of vehicle; (3) minimax distance criterion determination Optimal cluster center number k is adopted; (4) write KMeans clustering algorithm by MapReduce algorithm, Optimal cluster center number k is substituted in KMeans clustering algorithm, use the temporal characteristics vector of KMeans clustering algorithm to vehicle to carry out cluster analysis, obtain cluster result; (5) vehicle behavioural analysis is carried out according to cluster result.
Utilize the data in the distributed data base of Apache Pheonix middleware extraction intelligent traffic bayonet socket, carry out structuring to data, the data later to structuring set up index, accelerate search efficiency.Cross bayonet socket number of times according to MapReduce algorithm calculating vehicle each little period in one day, obtain the temporal characteristics vector of vehicle, i.e. 24 dimensional vector V
t.Distance between described cluster centre is the Euclidean distance in 24 dimension spaces between 2.
As shown in Figure 3, described minimax distance criterion comprises the following steps: (1) Schilling cluster centre number is 3, cluster is carried out to the temporal characteristics vector of vehicle, obtain 3 groups of clusters, average to often organizing cluster, this mean value is the cluster centre of this group cluster, the distance between the cluster centre calculating three groups of clusters, and MMD value when getting that wherein minor increment is 3 as cluster centre number; In like manner, try to achieve that cluster centre number is 4, MMD value corresponding to 5,6,7,8,9,10; (2) getting the maximum cluster centre number of MMD value is Optimal cluster center number k.Minimax distance criterion by compare adopt different cluster centre number cluster result in the spacing minimum value of all kinds of central point obtain Optimal cluster center number, the minimum value of the spacing of central point is larger, illustrates that the dispersion degree of each cluster centre is higher, then the effect of cluster is better.
Embodiment one
Determine cluster centre number: be first 3 to 10 carry out 7 clusters respectively with cluster centre number, obtain 7 groups of cluster centre groups, as follows:
(1) result of 3 classes is divided:
0 [1.0, 0.68, 0.53, 0.42, 0.46, 1.15, 2.69, 4.26, 8.48, 7.03, 6.09, 5.64, 4.56, 5.19, 6.05, 5.93, 6.23, 7.29, 6.1, 3.77, 3.21, 2.97, 2.21, 1.45]
1 [0.43, 0.22, 0.14, 0.13, 0.22, 0.65, 5.06, 36.89, 11.94, 3.95, 3.45, 4.09, 3.24, 3.82, 4.32, 3.8, 5.91, 15.72, 10.92, 4.31, 3.2, 2.65, 1.48, 0.73]
2 [6.31, 4.54, 3.5, 2.81, 2.72, 5.03, 10.52, 15.33, 23.08, 21.56, 19.74, 18.76, 16.0, 17.82, 19.72, 19.61, 20.07, 21.11, 16.83, 12.34, 11.03, 11.62, 10.77, 8.57]
(2) result of 4 classes is divided:
0 [5.81, 4.14, 3.16, 2.5, 2.3, 3.82, 7.46, 10.93, 17.58, 17.48, 16.08, 15.03, 13.06, 14.38, 15.89, 15.79, 15.31, 15.51, 13.25, 10.09, 9.59, 10.4, 9.91, 7.96]
1 [0.41, 0.21, 0.13, 0.12, 0.22, 0.59, 4.92, 40.0, 10.44, 3.81, 3.36, 4.12, 3.15, 3.79, 4.29, 3.68, 5.91, 15.68, 11.03, 4.34, 3.17, 2.69, 1.46, 0.69]
2 [3.88, 2.97, 2.32, 2.01, 2.78, 9.46, 23.77, 37.38, 50.54, 38.76, 34.45, 34.12, 26.99, 31.47, 35.62, 35.59, 42.16, 50.69, 34.48, 21.56, 14.68, 12.1, 8.43, 5.43]
3 [0.78, 0.52, 0.41, 0.34, 0.39, 1.04, 2.62, 4.41, 8.34, 6.4, 5.5, 5.1, 4.1, 4.68, 5.48, 5.36, 5.77, 7.09, 5.82, 3.44, 2.87, 2.58, 1.81, 1.13]
(3) result of 5 classes is divided:
0 [3.95, 2.73, 2.03, 1.56, 1.35, 2.3, 5.27, 9.35, 12.17, 14.37, 13.35, 12.54, 10.37, 11.59, 13.0, 12.81, 12.37, 12.89, 11.05, 8.07, 7.5, 7.8, 7.11, 5.51]
1 [0.44, 0.19, 0.11, 0.11, 0.22, 0.41, 1.27, 9.21, 41.45, 7.11, 4.02, 3.96, 3.59, 3.98, 4.77, 4.12, 5.44, 14.39, 10.19, 3.84, 2.87, 2.56, 1.47, 0.72]
2 [7.5, 5.56, 4.37, 3.61, 3.68, 7.33, 13.49, 16.02, 23.29, 24.95, 23.19, 21.85, 18.93, 21.09, 23.27, 23.75, 24.98, 23.93, 18.61, 14.2, 12.63, 13.45, 12.46, 9.96]
3 [0.68, 0.46, 0.37, 0.3, 0.37, 1.02, 2.86, 8.08, 5.75, 5.52, 4.85, 4.56, 3.68, 4.21, 4.91, 4.78, 5.3, 7.02, 5.72, 3.23, 2.66, 2.39, 1.65, 1.01]
4 [1.52, 0.98, 0.75, 0.77, 1.15, 4.1, 21.16, 82.63, 33.49, 21.55, 18.78, 21.22, 15.87, 19.14, 21.33, 19.09, 26.59, 49.41, 33.75, 17.69, 12.0, 8.48, 4.73, 2.42]
(4) result of 6 classes is divided:
0 [2.2, 1.48, 1.1, 0.79, 0.72, 1.55, 3.9, 6.64, 10.84, 13.67, 12.53, 11.63, 9.15, 10.35, 12.08, 12.12, 12.28, 12.74, 9.73, 6.34, 5.37, 5.17, 4.32, 3.13]
1 [0.43, 0.2, 0.11, 0.11, 0.18, 0.37, 1.17, 7.59, 44.1, 7.29, 4.07, 4.0, 3.74, 4.03, 4.87, 4.11, 5.39, 14.53, 10.5, 3.89, 2.87, 2.57, 1.5, 0.74]
2 [8.0, 5.82, 4.47, 3.64, 3.4, 5.83, 11.33, 14.6, 18.78, 19.92, 18.58, 17.44, 15.56, 17.24, 18.5, 18.44, 17.98, 18.19, 15.64, 12.55, 12.15, 13.47, 13.08, 10.72]
3 [0.77, 0.52, 0.42, 0.35, 0.42, 1.1, 2.74, 4.53, 5.82, 5.69, 4.96, 4.62, 3.76, 4.3, 4.99, 4.87, 5.21, 6.22, 5.35, 3.29, 2.8, 2.54, 1.81, 1.14]
4 [3.33, 2.45, 2.11, 1.82, 2.96, 10.24, 26.62, 47.71, 52.91, 48.73, 44.58, 44.03, 34.21, 39.06, 45.27, 45.45, 54.47, 60.32, 40.3, 26.65, 16.79, 12.81, 8.24, 4.92]
5 [0.4, 0.21, 0.13, 0.11, 0.22, 0.55, 4.84, 43.01, 8.1, 3.6, 3.28, 4.12, 3.12, 3.75, 4.23, 3.61, 5.9, 15.19, 10.93, 4.35, 3.23, 2.69, 1.43, 0.67]
(5) result of 7 classes is divided:
0 [2.73, 1.85, 1.34, 0.97, 0.84, 1.55, 3.6, 7.39, 12.02, 14.85, 13.64, 12.8, 10.13, 11.41, 13.26, 13.26, 13.09, 13.08, 9.97, 6.71, 5.87, 5.94, 5.21, 3.87]
1 [0.44, 0.21, 0.12, 0.12, 0.19, 0.38, 1.21, 7.74, 45.53, 7.28, 4.04, 4.01, 3.78, 4.06, 4.91, 4.11, 5.42, 15.08, 10.7, 3.97, 2.9, 2.66, 1.56, 0.75]
2 [8.3, 6.06, 4.7, 3.81, 3.66, 6.23, 12.03, 15.19, 19.21, 20.18, 18.85, 17.72, 15.85, 17.67, 18.8, 18.79, 18.49, 18.83, 16.25, 13.0, 12.53, 13.87, 13.51, 11.1]
3 [1.31, 0.91, 0.75, 0.69, 0.85, 2.59, 6.79, 7.66, 4.86, 3.05, 2.76, 2.94, 2.8, 3.08, 3.24, 3.21, 4.12, 8.77, 9.2, 5.82, 4.99, 4.38, 3.0, 1.87]
4 [3.42, 2.49, 2.19, 1.84, 2.68, 10.15, 26.46, 48.63, 52.92, 49.73, 45.58, 45.09, 35.14, 39.77, 46.54, 46.62, 55.92, 60.82, 41.02, 27.32, 17.17, 13.01, 8.45, 5.08]
5 [0.4, 0.2, 0.11, 0.1, 0.21, 0.49, 4.37, 45.23, 8.41, 3.7, 3.37, 4.28, 3.2, 3.89, 4.4, 3.72, 6.13, 15.56, 10.84, 4.31, 3.28, 2.68, 1.42, 0.67]
6 [0.51, 0.33, 0.25, 0.18, 0.19, 0.41, 1.08, 3.41, 6.68, 7.41, 6.42, 5.73, 4.44, 5.13, 6.14, 5.99, 6.11, 5.54, 3.86, 2.22, 1.82, 1.68, 1.23, 0.78]
(6) result of 8 classes is divided:
0 [0.68, 0.45, 0.34, 0.28, 0.3, 1.38, 5.62, 8.18, 9.42, 7.39, 6.75, 7.68, 5.8, 6.23, 7.73, 8.03, 11.49, 28.7, 11.75, 4.55, 3.2, 2.63, 1.68, 1.02]
1 [0.44, 0.22, 0.12, 0.12, 0.19, 0.4, 1.24, 7.31, 46.64, 7.59, 4.23, 4.17, 3.94, 4.24, 5.12, 4.28, 5.42, 13.57, 10.92, 4.14, 3.02, 2.71, 1.62, 0.76]
2 [6.74, 4.84, 3.7, 2.94, 2.72, 4.36, 8.27, 12.06, 16.67, 18.54, 17.34, 16.12, 14.13, 15.65, 16.99, 16.88, 16.01, 15.11, 13.43, 10.76, 10.48, 11.6, 11.27, 9.16]
3 [1.76, 1.21, 0.99, 0.86, 1.0, 2.76, 6.14, 4.8, 4.45, 3.31, 3.1, 3.26, 3.14, 3.5, 3.61, 3.53, 3.98, 5.72, 8.5, 6.49, 5.82, 5.28, 3.84, 2.49]
4 [1.3, 0.64, 0.42, 0.47, 0.6, 1.56, 12.95, 87.29, 21.09, 9.76, 8.68, 11.81, 8.09, 10.72, 10.95, 9.15, 16.01, 37.6, 25.24, 11.08, 8.8, 6.88, 3.19, 1.82]
5 [0.32, 0.18, 0.1, 0.08, 0.18, 0.42, 3.48, 34.01, 7.45, 3.29, 2.95, 3.46, 2.66, 3.26, 3.72, 3.22, 4.78, 9.89, 8.34, 3.33, 2.42, 2.11, 1.21, 0.55]
6 [0.49, 0.3, 0.23, 0.16, 0.18, 0.42, 1.19, 3.54, 7.21, 8.37, 7.25, 6.37, 4.87, 5.64, 6.79, 6.67, 6.66, 5.79, 4.15, 2.31, 1.81, 1.63, 1.16, 0.73]
7 [3.94, 3.01, 2.52, 2.07, 2.93, 11.27, 26.15, 34.3, 47.34, 48.16, 43.46, 41.56, 33.09, 37.35, 43.77, 44.74, 52.25, 53.74, 37.21, 24.92, 16.35, 12.81, 8.97, 5.59]
(7) result of 9 classes is divided:
0 [0.4, 0.28, 0.21, 0.19, 0.24, 1.68, 7.91, 9.61, 8.94, 4.61, 3.97, 5.14, 4.03, 4.12, 4.81, 4.61, 7.94, 33.67, 10.98, 3.36, 2.39, 1.96, 1.11, 0.64]
1 [0.46, 0.23, 0.12, 0.13, 0.2, 0.4, 1.27, 7.23, 47.54, 7.33, 4.07, 4.07, 3.94, 4.15, 5.03, 4.15, 5.42, 13.47, 11.16, 4.21, 3.05, 2.76, 1.64, 0.77]
2 [1.95, 1.26, 0.94, 0.68, 0.64, 1.31, 3.5, 8.19, 14.31, 18.12, 16.31, 14.97, 11.26, 13.12, 15.67, 15.65, 15.07, 13.12, 9.83, 6.24, 5.04, 4.82, 3.94, 2.88]
3 [1.94, 1.33, 1.07, 0.92, 1.06, 2.86, 6.11, 4.76, 4.51, 3.52, 3.3, 3.45, 3.29, 3.71, 3.82, 3.75, 4.17, 5.69, 8.94, 6.89, 6.25, 5.72, 4.19, 2.74]
4 [1.08, 0.61, 0.43, 0.44, 0.58, 1.53, 12.61, 86.0, 20.76, 9.93, 8.76, 11.81, 8.04, 10.63, 11.14, 9.16, 15.98, 37.09, 24.7, 10.95, 8.54, 6.63, 3.01, 1.65]
5 [0.32, 0.18, 0.1, 0.08, 0.19, 0.42, 3.43, 33.94, 7.42, 3.26, 2.93, 3.43, 2.67, 3.23, 3.68, 3.18, 4.77, 9.37, 8.41, 3.34, 2.45, 2.08, 1.2, 0.55]
6 [0.45, 0.28, 0.22, 0.16, 0.17, 0.4, 1.18, 3.48, 6.92, 7.62, 6.59, 5.86, 4.54, 5.23, 6.27, 6.15, 6.31, 5.78, 4.05, 2.21, 1.72, 1.56, 1.1, 0.67]
7 [3.49, 2.56, 2.24, 1.79, 2.47, 10.46, 25.26, 38.06, 53.0, 52.87, 48.14, 46.37, 36.24, 40.75, 47.71, 49.47, 57.64, 57.63, 40.33, 27.89, 17.52, 13.11, 8.72, 5.16]
8 [9.13, 6.66, 5.12, 4.14, 3.89, 6.31, 11.39, 14.01, 17.48, 18.1, 17.18, 16.06, 15.11, 16.36, 17.11, 17.13, 16.57, 16.92, 15.54, 13.1, 13.16, 14.87, 14.81, 12.2]
(8) result of 10 classes is divided:
0 [0.35, 0.25, 0.18, 0.17, 0.22, 1.33, 6.59, 9.93, 9.18, 4.34, 3.6, 4.75, 3.66, 3.78, 4.39, 4.21, 7.39, 33.42, 9.91, 2.91, 2.17, 1.76, 0.98, 0.56]
1 [0.46, 0.22, 0.12, 0.13, 0.21, 0.4, 1.23, 7.25, 47.89, 7.18, 3.98, 4.0, 3.83, 4.08, 4.99, 4.11, 5.37, 13.35, 11.21, 4.22, 3.04, 2.74, 1.6, 0.75]
2 [1.18, 0.77, 0.56, 0.41, 0.45, 1.16, 3.21, 7.45, 13.11, 17.01, 15.27, 13.95, 10.34, 11.96, 14.34, 14.25, 13.76, 12.05, 9.04, 5.52, 4.13, 3.66, 2.67, 1.78]
3 [1.75, 1.2, 0.99, 0.89, 1.07, 3.13, 6.84, 4.91, 4.38, 3.31, 3.07, 3.25, 3.1, 3.45, 3.53, 3.47, 3.97, 5.62, 9.25, 7.03, 6.23, 5.55, 3.87, 2.44]
4 [1.23, 0.58, 0.46, 0.47, 0.63, 1.5, 11.25, 89.25, 21.04, 9.73, 8.66, 11.4, 7.86, 10.39, 11.13, 9.17, 15.9, 37.91, 24.52, 10.67, 8.55, 6.72, 3.05, 1.8]
5 [0.3, 0.18, 0.1, 0.08, 0.19, 0.42, 3.51, 34.38, 7.45, 3.28, 2.94, 3.45, 2.68, 3.26, 3.73, 3.19, 4.81, 9.31, 8.54, 3.4, 2.45, 2.08, 1.19, 0.53]
6 [0.48, 0.31, 0.23, 0.17, 0.18, 0.39, 1.15, 3.47, 6.77, 7.26, 6.29, 5.62, 4.43, 5.07, 6.06, 5.95, 6.11, 5.66, 4.04, 2.23, 1.78, 1.65, 1.18, 0.73]
7 [5.0, 3.78, 2.96, 2.54, 3.03, 9.14, 19.26, 20.27, 26.4, 27.34, 25.3, 25.15, 21.25, 24.11, 25.99, 26.42, 30.12, 30.95, 23.53, 15.94, 12.27, 11.64, 9.35, 6.78]
8 [9.15, 6.55, 5.02, 3.94, 3.48, 4.43, 7.79, 11.53, 15.14, 16.08, 15.24, 13.84, 13.01, 14.1, 15.03, 15.03, 13.7, 13.64, 12.9, 11.5, 12.28, 14.33, 14.81, 12.37]
9 [3.05, 2.08, 1.59, 1.02, 1.03, 9.72, 28.99, 58.97, 83.8, 81.09, 73.08, 67.87, 50.19, 57.18, 70.52, 71.7, 80.44, 81.04, 58.85, 39.64, 24.97, 16.2, 10.13, 5.26]
The present invention proposes minimax distance criterion, namely calculate the Euclidean distance often organized in cluster centre group between each central point, and the minor increment in getting every group distance group compares.Distance group is as shown in following table table one:
Table one
As shown in Table 1, when cluster centre number gets 4, the Maximizing Minimum Distance between cluster centre point, according to minimax distance criterion, the dispersion degree between the cluster that this kind of cluster mode obtains is maximum, and Clustering Effect is ideal.Therefore, the present embodiment adopts cluster centre number to be 4 as Optimal cluster center number k.
When cluster numbers is 4, obtain cluster result, and taxi information contrasted as with reference to index, obtain following result:
Analyze cluster result, as shown in Figure 5:
In classification 1, total 12607, sample, wherein has 58.49% for taxi, it is characterized by: after when the morning 9 to 24 time trip comparatively average, when 0 to 8 time go out line frequency and reduce, but be not 0.Observe bayonet socket image to know, based on commerial vehicle in this type of result.
In classification 2, total 6072, sample, does not wherein almost have taxi, to it is characterized by 8 time and 18 time have obvious peak value, all the other times go out line frequency polar region, when 0 to 5 time do not go on a journey, observe bayonet socket image to know, this type of result is the private car, traffic regular bus etc. of commuter time rule.
In classification 3, total 1234, sample, sample number is minimum, and wherein taxi accounts for 15.32%, it is characterized by: when 9 and 18 time have obvious peak value, very high frequency of going on a journey daytime.Observe bayonet socket image to know, vehicle driving route upper latch notch is more thus embodies comparatively abnormal data for such.
In classification 4, total 40066, sample, wherein taxi ratio accounts for 3.16%, it is characterized by: when 9 and 18 time go out line frequency and slightly increase, when 8 to 22 time entirety to go out line frequency comparatively average, substantially without the situation of trip after when 24, such vehicle is without rule of obviously going on a journey.
In sum, core of the present invention is to ask for Optimal cluster center number k by minimax distance criterion, on the basis of Optimal cluster center number k, by calling MapReduce algorithm and KMeans clustering algorithm carries out cluster analysis to vehicle pass-through information, obtain cluster result.This method has validity and accuracy, by the division to vehicle, effectively can understand the behavioural characteristic of motor vehicle, carrying out cluster according to trip rule to vehicle can quick obtaining information of vehicles, for this area automobile traffic amount carries out the assessment of science, and provide reference frame to the prediction of future trajectory traffic passage situation, simultaneously also for public security deparment and traffic police department's investigation suspected vehicles provides reference frame.
Claims (5)
1., based on a motor vehicle clustering method for travel time feature, the method comprises the step of following order:
(1) from the distributed data base of intelligent traffic bayonet socket, extract the bayonet socket traffic information of vehicle;
(2) add up the number of times crossing bayonet socket of each little period of each car in one day, obtain the temporal characteristics vector of vehicle;
(3) minimax distance criterion determination Optimal cluster center number k is adopted;
(4) write KMeans clustering algorithm by MapReduce algorithm, Optimal cluster center number k is substituted in KMeans clustering algorithm, use the temporal characteristics vector of KMeans clustering algorithm to vehicle to carry out cluster analysis, obtain cluster result;
(5) vehicle behavioural analysis is carried out according to cluster result.
2. the motor vehicle clustering method based on travel time feature according to claim 1, it is characterized in that: utilize the data in the distributed data base of Apache Pheonix middleware extraction intelligent traffic bayonet socket, carry out structuring to data, the data later to structuring set up index.
3. the motor vehicle clustering method based on travel time feature according to claim 1, is characterized in that: cross bayonet socket number of times according to MapReduce algorithm calculating vehicle each little period in one day, obtains the temporal characteristics vector of vehicle, i.e. 24 dimensional vector V
t.
4. the motor vehicle clustering method based on travel time feature according to claim 1, is characterized in that: described minimax distance criterion comprises the following steps:
(1) Schilling cluster centre number is 3, cluster is carried out to the temporal characteristics vector of vehicle, obtain 3 groups of clusters, average to often organizing cluster, this mean value is the cluster centre of this group cluster, distance between the cluster centre calculating three groups of clusters, and MMD value when getting that wherein minor increment is 3 as cluster centre number; In like manner, try to achieve that cluster centre number is 4, MMD value corresponding to 5,6,7,8,9,10;
(2) getting the maximum cluster centre number of MMD value is Optimal cluster center number k.
5. the motor vehicle clustering method based on travel time feature according to claim 4, is characterized in that: the distance between described cluster centre is the Euclidean distance in 24 dimension spaces between 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510129468.5A CN104750800A (en) | 2014-11-13 | 2015-03-24 | Motor vehicle clustering method based on travel time characteristic |
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410638889 | 2014-11-13 | ||
| CN2014106388896 | 2014-11-13 | ||
| CN201510129468.5A CN104750800A (en) | 2014-11-13 | 2015-03-24 | Motor vehicle clustering method based on travel time characteristic |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104750800A true CN104750800A (en) | 2015-07-01 |
Family
ID=53590484
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510129468.5A Pending CN104750800A (en) | 2014-11-13 | 2015-03-24 | Motor vehicle clustering method based on travel time characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104750800A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105427602A (en) * | 2015-12-16 | 2016-03-23 | 浙江宇视科技有限公司 | Vehicle trip theme determining method and device |
| CN105718946A (en) * | 2016-01-20 | 2016-06-29 | 北京工业大学 | Passenger going-out behavior analysis method based on subway card-swiping data |
| CN105913656A (en) * | 2016-04-28 | 2016-08-31 | 泰华智慧产业集团股份有限公司 | Distributed statistics based method and system for frequently passing vehicles |
| CN106096507A (en) * | 2016-05-27 | 2016-11-09 | 中兴软创科技股份有限公司 | Wisdom traffic illegal vehicle recognition methods |
| WO2017092418A1 (en) * | 2015-12-04 | 2017-06-08 | 杭州海康威视数字技术股份有限公司 | Method of processing passage record and device |
| CN107292186A (en) * | 2016-03-31 | 2017-10-24 | 阿里巴巴集团控股有限公司 | A kind of model training method and device based on random forest |
| CN107862019A (en) * | 2017-10-31 | 2018-03-30 | 泰华智慧产业集团股份有限公司 | A kind of method and device for vehicle of being hidden by day and come out at night based on big data analysis |
| CN108389419A (en) * | 2018-03-02 | 2018-08-10 | 辽宁工业大学 | A kind of Dynamic Route Guidance Method of Vehicle |
| CN112257589A (en) * | 2020-10-22 | 2021-01-22 | 大连交通大学 | Method for detecting abnormally stopped vehicle |
| CN112963160A (en) * | 2021-02-10 | 2021-06-15 | 上海市隧道工程轨道交通设计研究院 | Method for identifying and determining stratum characteristics by using shield real-time tunneling parameters |
| CN116361679A (en) * | 2023-06-02 | 2023-06-30 | 青岛豪迈电缆集团有限公司 | Intelligent cable life prediction method and system based on data driving |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102184547A (en) * | 2011-03-28 | 2011-09-14 | 长安大学 | Video-based vehicle reverse driving event detecting method |
| CN102289659A (en) * | 2011-06-23 | 2011-12-21 | 杭州诚道科技有限公司 | Companion vehicle detection method |
| US20120330493A1 (en) * | 2011-06-24 | 2012-12-27 | Inter-University Research Institute Corporation, Research Organization of Information and System | Method and apparatus for determining road surface condition |
-
2015
- 2015-03-24 CN CN201510129468.5A patent/CN104750800A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102184547A (en) * | 2011-03-28 | 2011-09-14 | 长安大学 | Video-based vehicle reverse driving event detecting method |
| CN102289659A (en) * | 2011-06-23 | 2011-12-21 | 杭州诚道科技有限公司 | Companion vehicle detection method |
| US20120330493A1 (en) * | 2011-06-24 | 2012-12-27 | Inter-University Research Institute Corporation, Research Organization of Information and System | Method and apparatus for determining road surface condition |
Non-Patent Citations (2)
| Title |
|---|
| 张旭: "基于时空约束的轨迹聚类方法研究与应用", 《中国优秀硕士学位论文全文数据库》 * |
| 陈爱平: "基于Hadoop的聚类算法并行化分析及应用研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3385919A4 (en) * | 2015-12-04 | 2019-10-09 | Hangzhou Hikvision Digital Technology Co., Ltd. | Method of processing passage record and device |
| WO2017092418A1 (en) * | 2015-12-04 | 2017-06-08 | 杭州海康威视数字技术股份有限公司 | Method of processing passage record and device |
| US10810870B2 (en) | 2015-12-04 | 2020-10-20 | Hangzhou Hikvision Digital Technology Co., Ltd. | Method of processing passage record and device |
| CN105427602A (en) * | 2015-12-16 | 2016-03-23 | 浙江宇视科技有限公司 | Vehicle trip theme determining method and device |
| CN105718946A (en) * | 2016-01-20 | 2016-06-29 | 北京工业大学 | Passenger going-out behavior analysis method based on subway card-swiping data |
| US11276013B2 (en) | 2016-03-31 | 2022-03-15 | Alibaba Group Holding Limited | Method and apparatus for training model based on random forest |
| CN107292186A (en) * | 2016-03-31 | 2017-10-24 | 阿里巴巴集团控股有限公司 | A kind of model training method and device based on random forest |
| CN107292186B (en) * | 2016-03-31 | 2021-01-12 | 阿里巴巴集团控股有限公司 | Model training method and device based on random forest |
| CN105913656A (en) * | 2016-04-28 | 2016-08-31 | 泰华智慧产业集团股份有限公司 | Distributed statistics based method and system for frequently passing vehicles |
| CN106096507A (en) * | 2016-05-27 | 2016-11-09 | 中兴软创科技股份有限公司 | Wisdom traffic illegal vehicle recognition methods |
| CN106096507B (en) * | 2016-05-27 | 2020-03-24 | 浩鲸云计算科技股份有限公司 | Intelligent traffic black car identification method |
| CN107862019A (en) * | 2017-10-31 | 2018-03-30 | 泰华智慧产业集团股份有限公司 | A kind of method and device for vehicle of being hidden by day and come out at night based on big data analysis |
| CN108389419B (en) * | 2018-03-02 | 2024-03-29 | 辽宁工业大学 | Vehicle dynamic path induction method |
| CN108389419A (en) * | 2018-03-02 | 2018-08-10 | 辽宁工业大学 | A kind of Dynamic Route Guidance Method of Vehicle |
| CN112257589A (en) * | 2020-10-22 | 2021-01-22 | 大连交通大学 | Method for detecting abnormally stopped vehicle |
| CN112257589B (en) * | 2020-10-22 | 2024-03-05 | 大连交通大学 | Detection method for abnormal parking vehicle |
| CN112963160A (en) * | 2021-02-10 | 2021-06-15 | 上海市隧道工程轨道交通设计研究院 | Method for identifying and determining stratum characteristics by using shield real-time tunneling parameters |
| CN116361679A (en) * | 2023-06-02 | 2023-06-30 | 青岛豪迈电缆集团有限公司 | Intelligent cable life prediction method and system based on data driving |
| CN116361679B (en) * | 2023-06-02 | 2023-08-11 | 青岛豪迈电缆集团有限公司 | Intelligent cable life prediction method and system based on data driving |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104750800A (en) | Motor vehicle clustering method based on travel time characteristic | |
| CN105303197B (en) | A kind of vehicle follow the bus safety automation appraisal procedure based on machine learning | |
| CN106096507B (en) | Intelligent traffic black car identification method | |
| CN104794184B (en) | A kind of illegal vehicle recognition methods of the Bayesian Classification Arithmetic based on large-scale data | |
| CN108847022B (en) | Abnormal value detection method of microwave traffic data acquisition equipment | |
| CN107329977B (en) | A kind of false-trademark vehicle postsearch screening method based on probability distribution | |
| CN109118119A (en) | Air control model generating method and device | |
| CN102291392A (en) | Hybrid intrusion detection method based on bagging algorithm | |
| Aziz et al. | Identifying and characterizing truck stops from GPS data | |
| CN106297304A (en) | A kind of based on MapReduce towards the fake-licensed car recognition methods of extensive bayonet socket data | |
| CN112215497B (en) | Pure electric vehicle operation risk early warning method | |
| Chen et al. | Pattern recognition using clustering algorithm for scenario definition in traffic simulation-based decision support systems | |
| CN109767618B (en) | Comprehensive study and judgment method and system for abnormal data of public security traffic management service | |
| CN113205134A (en) | Network security situation prediction method and system | |
| CN117081858B (en) | Intrusion behavior detection method, system, equipment and medium based on multi-decision tree | |
| CN112652168B (en) | Major traffic accident early warning method, system and storage medium | |
| Li et al. | Enhancing driving safety: Discovering individualized hazardous driving scenes using GIS and mobile sensing | |
| CN114708728B (en) | Method for identifying traffic peak period, electronic equipment and storage medium | |
| Qiong et al. | Application of clustering algorithm in intelligent transportation data analysis | |
| CN113364881A (en) | Cloud service interaction analysis method applied to big data, server and storage medium | |
| CN112633163A (en) | Detection method for realizing illegal operation vehicle detection based on machine learning algorithm | |
| Sabarish et al. | Spatial Outlier Detection Algorithm for Trajectory-Data | |
| Chen et al. | Operational scenario definition in traffic simulation-based decision support systems: Pattern recognition using a clustering algorithm | |
| Li et al. | Research on abnormal monitoring of vehicle traffic network data based on support vector machine | |
| Milardo et al. | An unsupervised approach for driving behavior analysis of professional truck drivers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150701 |
|
| WD01 | Invention patent application deemed withdrawn after publication |