CN105336164A - Error checkpoint positional information automatic identification method based on big data analysis - Google Patents

Abstract

The invention discloses an error checkpoint positional information automatic identification method based on big data analysis. The method comprises the steps of: calculating adjacent checkpoint pairs according to mass historical checkpoint vehicle passing data, calculating the shortest path length of the adjacent checkpoint pairs by combining with GIS map data, and calculating passing speed of vehicles between the checkpoint pairs through vehicle data; and finding out the checkpoint with error positional information according to checkpoint shortest path abnormity, checkpoint interval velocity abnormity and abnormal frequency iteration in sequence. The error checkpoint positional information automatic identification method utilizes real checkpoint vehicle passing data, and has the advantages of large data size, high data accuracy and the like, thus the accuracy of error checkpoint positional information analyzed according to attributes of vehicle passing data is high. The error checkpoint positional information automatic identification method achieves the automatic identification of the error checkpoint positional information, has the advantages of fast execution speed, high accuracy, high efficiency and the like, and omits a great deal of manual identification.

Description

Based on the wrong bayonet socket positional information automatic identifying method of large data analysis

Technical field

The invention belongs to data mining technical field, be specifically related to a kind of automatic identifying method of the wrong bayonet socket positional information based on large data analysis.

Background technology

After development that is economic and society, particularly China 18 greatly, urbanization process is accelerated, city size constantly expands, urban population continues to increase, and city automobile recoverable amount rapidly increases, and particularly the automobile pollution of big city and megalopolis almost reaches peak value.Because the basic capacity of urban road is limited, cause traffic loading degree constantly to increase, traffic congestion is seen everywhere, and traffic problems have become a ubiquitous livelihood issues.

Along with the proposition of " internet+", wisdom traffic is also pushed to a new climax.Bayonet socket data, as one of the important component part of traffic data, are widely applied in the calculating of Real-time Road passage rate, traffic load calculating, vehicle behavior patterns mining etc.To finding during existing bayonet socket data research experiment that the positional information of bayonet socket is very crucial basic data, once these positional informations are made mistakes, the data mining of car information is crossed and large data analysis will produce serious misleading consequence based on bayonet socket, have a strong impact on correctness and the reliability of data analysis, cause for bayonet socket data analysis and research can not carry out smoothly, due to bayonet socket One's name is legion, nearly tens thousand of, and be distributed in each corner in city, if by manually examining comparison one by one, by the artificial of at substantial and transport cost, and the bayonet socket positional information of major part mistake is difficult to by manually going to find and get rid of.

How by the means of data analysis, in tens thousand of bayonet socket positional informations, those bayonet sockets that location information errors suspicion is larger can be identified automatically, then examine comparison by manual site, will greatly reduce artificial expense and time.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, provide a kind of wrong bayonet socket positional information robotization recognition methods based on large data analysis.

In the present invention, so-called bayonet socket refers to the road monitoring point being provided with traffic monitoring apparatus, for crossing the collection of car data, cross car data and refer to vehicle through bayonet socket, by the information of vehicles that bayonet socket collects, comprise " number-plate number " of vehicle, " bayonet socket numbering " and " spending the car time "; Adjacent bayonet socket is to the bayonet socket pair referring to that physical location is adjacent, can be sequentially passed through successively by vehicle.

The central scope of technical solution of the present invention: the history bayonet socket according to magnanimity crosses car data, calculates adjacent bayonet socket pair, calculates the right shortest path length of adjacent bayonet socket in conjunction with GIS map data, and calculates vehicle passage rate between bayonet socket pair according to crossing car data.The bayonet socket that between, bayonet socket abnormal to shortest path according to bayonet socket pair, overall speed is abnormal, abnormal frequency iteration finds out positional information mistake successively.

The inventive method comprises the following steps:

Step (1). import bayonet socket data in a database, bayonet socket crosses car data, map datum, bayonet socket packet has bayonet socket to number (KKBH) containing field, longitude (JD), latitude (WD), bayonet socket is crossed car data and is comprised field and have brand number (HPHM), bayonet socket numbering (KKBH), spend the car time (GCSJ), map datum comprises field ID (section major key), initial longitude and latitude (first_JD, first_WD), intermediate point longitude and latitude (center_JD, center_WD), terminal longitude and latitude (end_JD, end_WD).

Step (2). setting threshold value, had car amount threshold value GCL_max, path length threshold LJCD_max, large threshold speed DSD_max, little threshold speed XSD_min, velocity sag rate threshold value SDYCL_max, suspicion bayonet socket threshold X YKK_max respectively.

Step (3). the bayonet socket reading a period of time crosses car data, if the car amount of crossing between two adjacent bayonet sockets is greater than bayonet socket cross vehicle threshold value GCL_max, then these two bayonet sockets are defined as adjacent bayonet socket pair.

Step (4). the positional information right according to adjacent bayonet socket goes out shortest path length along GIS map path computing.

Step (5) if. the right path of the adjacent bayonet socket that step (4) calculates is greater than path length threshold LJCD_max, then by this bayonet socket to regarding as suspicion bayonet socket pair, and its information to be stored in database.

Step (6). the bayonet socket reading a period of time crosses car data, the path right according to the adjacent bayonet socket crossing car mistiming and process crossing car data calculates this car through the right speed of adjacent bayonet socket, judge whether speed is greater than large threshold speed DSD_max or little threshold speed XSD_min, and record its quantity and total car amount excessively.

Step (7) if. velocity sag quantity accounted for always to be crossed vehicle and is greater than velocity sag rate threshold value SDYCL_max, assert that bayonet socket is to being suspicion bayonet socket pair for this reason, and its information is stored in database.

Step (8). repeat step (5), step (6) for several times; Suspicion bayonet socket in staqtistical data base, draws the frequency of identical suspicion bayonet socket, if the frequency is greater than suspicion bayonet socket threshold X YKK_max, then assert that this bayonet socket is the bayonet socket of errors present information.

The beneficial effect that the present invention has: the present invention utilizes real bayonet socket to cross car data, has that data volume is large, data accuracy high, high according to the accuracy of the attributive analysis wrong bayonet socket positional information out crossing car data thus.Present invention achieves robotization identification error bayonet socket positional information, and there is the features such as execution speed is fast, accuracy is high, high-level efficiency.Avoid a large amount of artificial examination work.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of identification error positional information bayonet socket.

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.As shown in Figure 1, the present invention includes following steps:

Step (1). data encasement: import bayonet socket data in a database, bayonet socket crosses car data, map datum, bayonet socket packet has bayonet socket to number (KKBH) containing field, longitude (JD), latitude (WD), bayonet socket is crossed car data and is comprised field and have brand number (HPHM), bayonet socket numbering (KKBH), spend the car time (GCSJ), map datum comprises field ID (section major key), initial longitude and latitude (first_JD, first_WD), intermediate point longitude and latitude (center_JD, center_WD), terminal longitude and latitude (end_JD, end_WD).

Step (2). train adjacent bayonet socket to (training data is the car data excessively of 1-3 hour):

2-1. initialization map grid: map is divided square node by certain length of side (arranging the length of side is 100-200 rice), if the mid point latitude and longitude coordinates of certain road is in certain square mesh, then think that this road is arranged in this mesh, this road ID is put in corresponding " mesh-road mapping table " (as shown in table 3) of this mesh, key is mesh numbering, and value is that road ID gathers; If do not have corresponding " mesh-road mapping table ", be then newly-built empty " mesh-road mapping table " of this mesh, and put into this road ID.

Table 3

2-2. creates bayonet socket to mapping table map (as shown in table 2), and key is that bayonet socket is to numbering (A bayonet socket is numbered, and B bayonet socket is numbered), value { crosses car amount, path, large speed number, little speed number, suspicion identifies }, map is designated as { K1, K2}{gcl, ljcd, dsd, xsd, flag}.

Table 2

2-3. sets threshold value, has car amount threshold value GCL_max respectively, path length threshold LJCD_max, large threshold speed DSD_max, little threshold speed XSD_min, velocity sag rate threshold value SDYCL_max, suspicion bayonet socket threshold X YKK_max (as shown in table 4).

Table 4

2-4. is by " number-plate number ", the ascending order " spending the car time " reads bayonet socket and crosses car data (training data is generally greater than 1 hour), and each is recorded as { " license plate number ", " bayonet socket numbering ", " spend the car time " }, be designated as---cross car record { h, K, t}, wherein h representative " license plate number ", K represents " bayonet socket numbering ", and t representative " spends the car time ".

2-5. reads next and crosses car data, current record is { h2, K2, t2}, upper one is recorded as { h1, K1, t1}, if h2==h1, then searching bayonet socket to whether there is key in mapping table map is { K1, the record of K2}, if do not exist, newly-built { K1 in bayonet socket is to mapping table map, the record of K2} and the initial value of value is set to 0 entirely, if exist, by bayonet socket to { K1 in mapping table map, gcl in K2} record adds 1, judge whether gcl was greater than car amount threshold value GCL_max, if be greater than, skip to step 2-6, if be not more than, repeat step 2-5 until the reading of all car datas is excessively complete skip to step 2-7.

2-6. calculating path length: according to { K1, K2} two bayonet socket latitude and longitude coordinates, calculate the square node that two bayonet sockets are positioned at respectively, all road ID are drawn according to " mesh-road mapping table " in 2-1, calculate the distance of bayonet socket and each road, find out apart from minimum road, then think that this bayonet socket is positioned at this road, then with two bayonet socket place square nodes for diagonal angle end points, determine " the current rectangular area " of two bayonet sockets, calculate the length in this path according to the connectivity in the section in " current rectangular area ".

2-7. finally obtains adjacent bayonet socket to set map.

Step (3). identification error bayonet socket (experimental data, the data of general 5-10 minute):

The gcl of records all in map is set to 0 (needing again to add up gcl after changing experimental data) by 3-1., by " number-plate number ", the ascending order " spending the car time " reads bayonet socket and crosses car data, read next and cross car data, current record is { h2, K2, t2}, upper one is recorded as { h1, K1, t1}, if h2==h1, then search bayonet socket to whether there is key in mapping table map for { record of K1, K2}, if exist gcl+1 and calculate vehicle in the right travel speed of this adjacent bayonet socket according to formula (1).

sd＝ljcd/|t ₂-t ₁|(1)

Wherein sd represent bayonet socket the overall speed between K1, K2}, and ljcd be bayonet socket to the field of in mapping table map, meaning and bayonet socket { shortest path between K1, K2}, t ₂, t ₁be in car record vehicle through the time of bayonet socket K1, K2.

Judge whether sd is greater than large threshold speed DSD_max, if sd>DSD_max, then in map, the dsd of this record adds 1, otherwise judge whether sd is less than little threshold speed XSD_min, if sd<XSD_min, then in map, the xsd of this record adds 1, otherwise next crosses car data until all data all process to continue reading.

3-2. reads the record in map in order, judge whether the path of this record is greater than path length threshold, if ljcd>LJCD_max, suspicion is identified flag and is set to 1, calculate large speed in frequency dr respectively as shown in formula (2)

$dr=\frac{dsd}{gcl}*\frac{1}{{\log }_2\left(\frac{gcl+GCL}{gcl}\right)}---\left(2\right)$

The frequency xr of little speed as shown in formula (3),

$xr=\frac{xsd}{gcl}*\frac{1}{{\log }_2\left(\frac{gcl+GCL}{gcl}\right)}---\left(3\right)$

(GCL for this reason period cross car amount threshold value), if dr>SDYCL_max or xr>SDYCL_max, is then set to 1 by the suspicion of this record mark flag.

3-3. repeats 3-2 until all records are all disposed in map, to be stored in by all records of flag>0 in map in database in tables of data BAYONET_PAIR.

Step (4). that selectes different time sections crosses car data record, and by the dsd of records all in map, xsd, falg are set to 0, repeats step (3).

Step (5). the frequency of the identical bayonet socket in statistics table BAYONET_PAIR, if the frequency is greater than suspicion bayonet socket threshold X YKK_max, then this bayonet socket is regarded as the bayonet socket of positional information mistake.

Claims (1)

1., based on the automatic identifying method of the wrong bayonet socket positional information of large data analysis, it is characterized in that the concrete steps of the method are:

Step (1). importing bayonet socket data, bayonet socket cross car data, map datum in a database, bayonet socket packet has bayonet socket numbering KKBH, longitude JD, latitude WD containing field, bayonet socket cross car data comprise field have brand number HPHM, bayonet socket numbering KKBH, cross car time GCSJ, map datum comprises field section major key, initial longitude and latitude (first_JD, first_WD), intermediate point longitude and latitude (center_JD, center_WD), terminal longitude and latitude (end_JD, end_WD);

Step (2). setting threshold value, had car amount threshold value GCL_max, path length threshold LJCD_max, large threshold speed DSD_max, little threshold speed XSD_min, velocity sag rate threshold value SDYCL_max, suspicion bayonet socket threshold X YKK_max respectively;

Step (3). the bayonet socket reading a period of time crosses car data, if the car amount of crossing between two adjacent bayonet sockets is greater than bayonet socket cross vehicle threshold value GCL_max, then these two bayonet sockets are defined as adjacent bayonet socket pair;

Step (4). the positional information right according to adjacent bayonet socket goes out shortest path length along GIS map path computing;

Step (5) if. the right path of the adjacent bayonet socket that step (4) calculates is greater than path length threshold LJCD_max, then by this bayonet socket to regarding as suspicion bayonet socket pair, and its information is stored in database;

Step (6). the bayonet socket reading a period of time crosses car data, the path right according to the adjacent bayonet socket crossing car mistiming and process crossing car data calculates this car through the right speed of adjacent bayonet socket, judge whether speed is greater than large threshold speed DSD_max or little threshold speed XSD_min, and record its quantity and total car amount excessively;

Step (7) if. velocity sag quantity accounted for always to be crossed vehicle and is greater than velocity sag rate threshold value SDYCL_max, assert that bayonet socket is to being suspicion bayonet socket pair for this reason, and is stored in database by its information;

Step (8). repeat step (5), step (6) for several times; Suspicion bayonet socket in staqtistical data base, draws the frequency of identical suspicion bayonet socket, if the frequency is greater than suspicion bayonet socket threshold X YKK_max, then assert that this bayonet socket is the bayonet socket of errors present information.