CN105575120A - Floating car data parking behavior mode cleaning method specific to road real time speed calculation - Google Patents

Abstract

Provided is a floating car data parking behavior mode cleaning method specific to road real time speed calculation, comprising the steps of: (1) reading map data which comprises road section numbers and road section point location information, and gridding a map to obtain a ''grid-road section number'' mapping table; (2) reading in an ascending sequence the floating car data of a preset time slice before the current time according to ''license plate number'' and ''creation time''; (3) calculating the vehicle distance and time difference between twice continuous recordings, calculating the speed according to the distance and time difference, and adding the speed to the road section number where the vehicle point location information of secondary recording is located according to the ''grid-road section number'' mapping table; (4) obtaining the speed list of each vehicle at each road section of the time slice; and (5) identifying and rejecting abnormal stop points in the speed list caused by ''pseudo stop behaviors'', and returning to the speed list. The floating car data parking behavior mode cleaning method can effectively identify floating car parking behaviors.

Description

Towards the floating car data parking behavior pattern cleaning method that road real-time speed calculates

Technical field

The present invention relates in data mining technical field, be specifically related to a kind of floating car data parking behavior pattern cleaning method calculated towards road real-time speed.

Background technology

Road speeds is point duty, navigation, induced travel, block up administer etc. macroscopic view, microcosmic application prerequisite, be important basic data.Therefore obtain more close to real data

Floating car data, as one of the important component part of traffic data, has calculated at Real-time Road passage rate and has applied widely.Can find during existing floating car data research experiment, the data that the Floating Car behavior that abends produces are applied to the accuracy that can have a strong impact on its speed during Real-time Road passage rate calculates.Because floating car data amount is large, and the behavior individual behaviour and irregularly to follow of abending, artificial being difficult to finds timely and rejects.

Summary of the invention

In order to overcome the existing deficiency that effectively cannot identify parking behavior, the invention provides a kind of floating car data parking behavior pattern cleaning method calculated towards road real-time speed of effective identification Floating Car parking behavior.

The technical solution adopted for the present invention to solve the technical problems is:

Towards the floating car data parking behavior pattern cleaning method that road real-time speed calculates, described cleaning method comprises the steps:

Step (1). read map datum, map datum comprises section numbering (LDID), some position, section information, map grid is obtained " grid-section numbering " mapping table;

Step (2). according to the floating car data of setting-up time sheet before " license plate number ", " creation-time " ascending order reading current time, floating car data comprises number-plate number CPHM, some position information GPS_FDC and creation-time CJSJ;

Step (3). calculate the Distance geometry mistiming of the double record of vehicle, then by Distance geometry mistiming computing velocity, and according to " grid-section numbering " mapping table, this speed is added in the section numbering at the information place, vehicle point position of second time record;

Step (4). the speed list of each vehicle on each section of this timeslice is obtained according to above step;

Step (5). identify and reject abending a little of causing due to " pseudo-stopping behavior " in speed list, and return speed list, process is as follows:

In the data structure LDID_SpeedMap that 5-1. obtains in step (4), searching loop obtains CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 and v _i+s≠ 0,0<s<=n-i, s are the number of continuous halt, then the car speed record number δ=s+1 of " pseudo-stopping behavior " process;

5-2. calculates " pseudo-stopping behavior " car speed sudden change threshold value SPSSC according to formula (2); When i ≠ 1, the formula left side is the average velocity of all speed records before this speed list first halt, as i=1, with the average velocity v in a moment on this section _tas average velocity threshold value;

$SPSSC=\left\{\begin{array}{cc}\frac{\underset{f=1}{\overset{i-1}{\&Sigma;}}{v}_f}{i-1}*\mathrm{\&delta;},& i\&NotEqual;1\\ v_t*\mathrm{\&delta;},& i=1\end{array}\right\}---\left(2\right)$

If 5-3. is v _i+s>=SPSSC, then this speed record is the mutating speed that " pseudo-stopping behavior " causes, and rejects speed record and the mutating speed speed record of continuous halt from speed list;

5-4. completes " pseudo-stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

Further, described cleaning method also comprises the steps:

Step (6). identify and reject abending a little of causing due to " really stopping behavior " in speed list, and return speed list, process is as follows:

6-1. after step (5) from new LDID_SpeedMap searching loop obtain CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 and v _i+s≠ 0,0<s<=n-i, s are the number of continuous halt; If s>=3, think that this car has occurred that " stopping behavior " skips to 6-2 and judge whether to belong to " crossing stopping behavior ";

6-2. obtains the v of this stopping behavior _icorresponding LDID and GPS_FDC _i, traversal intersectionMap, if there is this LDID in intersectionMap, illustrates that this LDID is section, crossing, enters 6-3 if do not exist; From intersectionMap, obtain crossing point (GPS_QD, GPS_ZD) that this LDID is corresponding, calculate GPS_FDC by formula (3) _iget its minimum value L with the distance at crossing, if L<L_min, then think that this " stopping behavior " belongs to " crossing stopping behavior " and do not reject, otherwise enter 6-3 and judge whether to belong to " collective stops behavior ";

L＝min(abs(GPS_FDC _i-GPS_QD),abs(GPS_FDC _i-GPS_ZD)(3)

6-3. obtains the v of this stopping behavior _icorresponding CJSJ _iand GPS_FDC _iif the speed list of other vehicles occurs that " stopping behavior " obtains the GPS_FDC of its corresponding halt under this LDID of searching loop _i'and CJSJ _i', calculate Rule of judgment according to formula (4), formula (5),

abs(GPS_FDC _i-GPS_FDC _i')＜Δs_min,i≠i'(4)

abs(CJSJ _i-CJSJ _i')＜Δt_min,i≠i'(5)

If other vehicles all meet above Rule of judgment under this LDID, " collective stops behavior " does not reject to think that this stopping behavior belonging to, otherwise this stopping behavior belonging to " really stopping behavior ", and rejects the speed record of continuous halt from speed list;

6-4. completes " really stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

Further again, described cleaning method also comprises the steps:

Step (7). after above step, calculate the average velocity of all speed records corresponding to LDID and be stored in database real-time speed table, update time, sheet repeated step (2) to step (6).

Further, in described step (3), velocity computation process is as follows:

3-1. obtains all records of each vehicle under this timeslice according to step (2), is designated as { CPHM _k} _ { GPS_FDC _i, CJSJ _i, k ∈ Ki ∈ n, n is the number of all records, and K is the sum of this timeslice vehicle;

3-2. calculates car speed v according to formula (1.1) or (1.2) _i, obtain car speed list { CPHM _k} _ { v _i, GPS_FDC _i, CJSJ _i.

What adopt is the absolute distance of the GPS point of double record, as the operating range of vehicle,

1) when not across using formula (1.1) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.1\right)$

2) when across using formula (1.2) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{JD}}_i)+abs(GPS\_{\mathrm{JD}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.2\right)$

GPS_JD _ifor the GPS point of double record, when across section, the some position information of the intersection point in two sections.

In described step (7), obtain the speed record of all car speed lists under LDID according to LDID_SpeedMap and calculate average velocity corresponding to LDID according to formula (6), and applying for other stored in database Real-time Road velometer.

$v_{avg}=\frac{\underset{k=1}{\overset{K}{\&Sigma;}}\underset{i=1}{\overset{n_k}{\&Sigma;}}{v}_i}{K\&CenterDot;n_k}---\left(6\right)$

K is the vehicle number under this LDID, n _kfor the speed record number of a kth car.

Technical conceive of the present invention is: according to real-time floating car data, calculates floating vehicle travelling speed and adds in corresponding section, obtains the speed list of each vehicle in section and identifies the behavior that abends and rejecting abnormalities speed.

Described Floating Car refers to taxi in urban public transport and bus.Floating Car (can be generally 30 seconds) at set intervals and upload a secondary data in vehicle travel process, and data comprise: Floating Car license plate number, Floating Car positional information (GPS), data creation time etc.; " stopping behavior " comprises " normally stopping behavior " and " abend behavior "." normally stop behavior " to comprise:

" crossing stopping behavior ": vehicle is parked in crossing (in most cases waiting for that red light belongs to the normal phenomenon in traffic process)

" collective stop behavior ": refer to the behavior (there will be the normal phenomenon also belonged in traffic process when running into and blocking up) that stops of same place at one time of all Floating Car.

" abend behavior " comprising:

" really stop behavior ": vehicle abnormality stops (individual vehicle is parked in a point for a long time, and in traffic process, belong to individual behaviour is abnormal occurrence)

" pseudo-stopping behavior ": the abnormal stopping caused of data upload, shows as and occur that speed is the point of 0 continuously, and velocity jump after this point.(as shown in Figure 1).

" really stopping behavior ", " pseudo-stopping behavior " is distinguished from the feature of data itself." really stop behavior " and comprise " true stopping _ individual behavior " and " true stopping _ group behavior ".This programme identification be " true stopping _ individual behavior ", the individual behavior of this stopping is disturbing factor when calculating road speeds, will not count sample data when calculating road real-time speed.For " pseudo-stopping behavior " that this patent is set forth, mainly refer to that those are because the looking of causing such as equipment failure, transmission network fault or database Write fault is as the data sample stopping behavior.Although these stopping behaviors showing as stopping on data mode, the possibility that data are produced and transmission link mistake causes is larger, and in other words, we think that it is " pseudo-stopping behavior ".This " pseudo-stopping behavior " is also got rid of outside the sample data calculating road real-time speed.

Beneficial effect of the present invention is mainly manifested in: two kinds of stopping behaviors at big data quantity and in irregular governed floating car data fast and accurately in identification data are also rejected, while the real-time ensureing data, significantly increase the accuracy of real-time speed.

Accompanying drawing explanation

Fig. 1 is " pseudo-stopping behavior " key diagram.

Fig. 2 is the operating range calculating schematic diagram across vehicle when section.

Fig. 3 is the process flow diagram of the floating car data parking behavior pattern cleaning method towards the calculating of road real-time speed.

Fig. 4 is LDID_SpeedMap data structure diagram.

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.

With reference to Fig. 1 ~ Fig. 4, a kind of floating car data parking behavior pattern cleaning method calculated towards road real-time speed, definition related symbol is as follows:

V _i: car speed, unit km/h, i ∈ n, n are the speed record number of certain car on same LDID under single timeslice.

V _avg: Road average-speed, unit km/h.

δ: the car speed record number of " pseudo-stopping behavior " process (speed be continuously 0 number of times add 1 velocity jump).

SPSSC (StopPointSpeedSuddenChange): " pseudo-stopping behavior " car speed sudden change threshold value.

Δ t_min: the speed record mistiming threshold value (establishing value 30 seconds) of same section different vehicle.

Δ s_min: the speed record location interval threshold value (arranging 50 meters) of same section different vehicle

L_min: judge whether vehicle is parked in the distance threshold (arranging 50 meters) at crossing.

Described floating car data parking behavior pattern cleaning method comprises the steps:

Step (1). map datum pre-service:

1-1. reads map datum, map datum comprises section numbering (LDID), starting point point position information (GPS_QD), terminal point position information (GPS_ZD), mid point point position information (GPS_MD), map grid (grid length of side 100-300 rice) is obtained " grid-section numbering " mapping table (as shown in table 1).

Table 1

1-2. is because the connectivity in section, and between section 1 with section 2, terminal is connected with starting point (being same point), and in statistical map data, the starting point in all sections and terminal, obtain " point-section is numbered " mapping table (as shown in table 2).

Table 2

1-3. finds the point of fork in the road or crossroad and LDID number to be more than or equal to the point of 3 according to " point-section numbering " mapping table, form new " point-section numbering " mapping table to be called " crossing point mapping table ", " crossing point mapping table " is carried out upset formation " section numbering-point " mapping table (as shown in table 3, note: what record in Value is the some position information that this section belongs to fork in the road or crossroad, if the starting point in certain section and terminal all belong to fork in the road or four crossway stomion, in Value, record starting point point position information and the terminal point position information in this section, if only have wherein some points to be fork in the road or four crossway stomion, then only record the some position information of this point) be called that " section, crossing mapping table " remembers intersectionMap.

Table 3

Step (2). read the floating car data of current time toward previous timeslice (5min) according to " number-plate number ", " creation-time " ascending order, floating car data comprises the number-plate number (CPHM), some position information (GPS_FDC), creation-time (CJSJ).

Step (3). calculate road speeds:

3-1. obtains all records of each vehicle under this timeslice according to step (2), is designated as { CPHM _k} _ { GPS_FDC _i, CJSJ _i, k ∈ Ki ∈ n, n is the number of all records, and K is the sum of this timeslice vehicle.

3-2. calculates car speed v according to formula (1.1) or (1.2) _i, obtain car speed list { CPHM _k} _ { v _i, GPS_FDC _i, CJSJ _i.

What adopt is the absolute distance of the GPS point of double record, as the operating range of vehicle.

1) when not across using formula (1.1) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.1\right)$

2) when across using formula (1.2) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{JD}}_i)+abs(GPS\_{\mathrm{JD}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.2\right)$

GPS_JD _ifor the GPS point of double record, when across section, the some position information of the intersection point in two sections.

Step (4). according to the GPS_FDC in speed list _icoupling " grid-section numbering " mapping table, obtains GPS_FDC _iall LDID of affiliated grid, calculate GPS_FDC _icalculate distance with the GPS_MD of each LDID, get apart from the section of minimum LDID as this Point matching, and this speed list is added in LDID, obtain data structure LDID_SpeedMap (as shown in Figure 3).Step (5). identify and reject " pseudo-stopping behavior ":

5-1. searching loop in data structure LDID_SpeedMap obtains CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 (i.e. halt) and v _i+s≠ 0 (0<s<=n-i, s are the number of continuous halt), then the car speed record number δ=s+1 of " pseudo-stopping behavior " process.

5-2. calculates " pseudo-stopping behavior " car speed sudden change threshold value SPSSC according to formula (2).When i ≠ 1, the formula left side is the average velocity of all speed records before this speed list first halt, as i=1, with the average velocity v in a moment on this section _tas average velocity threshold value.

$SPSSC=\left\{\begin{array}{cc}\frac{\underset{f=1}{\overset{i-1}{\&Sigma;}}{v}_f}{i-1}*\mathrm{\&delta;},& i\&NotEqual;1\\ v_t*\mathrm{\&delta;},& i=1\end{array}\right\}---\left(2\right)$

If 5-3. is v _i+s>=SPSSC, then this speed record is the mutating speed that " pseudo-stopping behavior " causes, and rejects speed record and the mutating speed speed record of continuous halt from speed list.

5-4. completes " pseudo-stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

Citing:

In data structure LDID_SpeedMap, obtain the speed list of certain section car, be (35.52,27.89,0,0,131.07,32.12,9.1,33.17) in chronological order, unit is km/h.Now we can find that continuous halt number is 2 then " pseudo-stopping behavior " SPSSC (velocity jump threshold value)=(35.52+27.89)/2* (2+1)=31.7*3=95.1; 95.1<131.07 then thinks that the behavior is " pseudo-stopping behavior ".

Step (6). identify and reject " really stopping behavior "

Citing: " really stopping behavior ", as shown in the table: in same section, sheet only has a car to occur parking behavior at one time.Namely the behavior is " really stopping behavior "

Zhejiang CT1**0	Speed	Zhejiang CT3**3	Speed	Zhejiang CT3**7	Speed	Zhejiang CT5**6	Speed
								8:55:45	12.29	8:55:24	0	8:55:35	34.4	8:55:49	55.5
8:56:03	42.26	8:55:36	0	8:56:07	24.33	8:56:15	28.56
								8:56:25	27.62	8:55:57	4.93	8:56:31	85.7	8:56:51	0
8:56:44	8.29	8:56:18	11.14	8:57:02	25.41	8:57:14	36.2
								8:56:44	8.29	8:56:37	19.19	8:57:32	75.01	8:57:50	32.5
8:57:04	47.48	8:56:55	8	8:58:03	21.49	8:58:14	47.82
								8:57:25	50.36	8:56:55	8	8:58:32	0	8:58:50	22.25
8:57:43	50.5	8:57:17	0	8:58:56	11.25	8:59:15	28.98
								8:58:04	21	8:57:36	18.24	8:59:23	54.17	8:59:48	18
8:58:24	27	8:57:55	0	8:59:23	54.46
								8:58:45	25.07	8:58:17	0
8:59:04	22.03	8:58:37	2.7
								8:59:25	4.29	8:58:56	0
8:59:45	19.35	8:59:18	12.27

6-1. after step (5) from new LDID_SpeedMap searching loop obtain CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 (i.e. halt) and v _i+s≠ 0 (0<s<=n-i, s are the number of continuous halt); If s>=3, think that this car has occurred that " stopping behavior " skips to 6-2 and judge whether to belong to " crossing stopping behavior ".

6-2. obtains the v of this stopping behavior _icorresponding LDID and GPS_FDC _i.Traversal intersectionMap, if there is this LDID in intersectionMap, illustrates that this LDID is section, crossing, enters 6-3 if do not exist.From intersectionMap, obtain crossing point (GPS_QD, GPS_ZD) that this LDID is corresponding, calculate GPS_FDC by formula (3) _iget its minimum value L with the distance at crossing, if L<L_min, then think that this " stopping behavior " belongs to " crossing stopping behavior " and do not reject, otherwise enter 6-3 and judge whether to belong to " collective stops behavior ".

L＝min(abs(GPS_FDC _i-GPS_QD),abs(GPS_FDC _i-GPS_ZD)(3)

6-3. obtains the v of this stopping behavior _icorresponding CJSJ _iand GPS_FDC _iif the speed list of other vehicles occurs that " stopping behavior " obtains the GPS_FDC of its corresponding halt under this LDID of searching loop _i'and CJSJ _i', calculate Rule of judgment according to formula (4), formula (5),

abs(GPS_FDC _i-GPS_FDC _i')＜Δs_min,i≠i'(4)

Abs (CJSJ _i-CJSJ _i') < Δ t_min, i ≠ i'(5) " collective stops behavior " does not reject if other vehicles all meet above Rule of judgment under this LDID, to think that this stopping behavior belonging to, otherwise this stopping behavior belonging to " really stopping behavior ", and rejects the speed record of continuous halt from speed list.

6-4. completes " really stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

Step (7). after processing all data of this timeslice, obtain the speed record of all car speed lists under LDID according to LDID_SpeedMap and calculate average velocity corresponding to LDID according to formula (6), and applying for other stored in database Real-time Road velometer.

$v_{avg}=\frac{\underset{k=1}{\overset{K}{\&Sigma;}}\underset{i=1}{\overset{n_k}{\&Sigma;}}{v}_i}{K\&CenterDot;n_k}---\left(6\right)$

K is the vehicle number under this LDID, n _kfor the speed record number of a kth car.

Update time, sheet, repeated step (2) to step (6).

Claims (5)

1., towards the floating car data parking behavior pattern cleaning method that road real-time speed calculates, it is characterized in that: described cleaning method comprises the steps:

Step (1). read map datum, map datum comprises section numbering LDID, some position, section information, map grid is obtained " grid-section numbering " mapping table;

Step (2). according to the floating car data of setting-up time sheet before " license plate number ", " creation-time " ascending order reading current time, floating car data comprises number-plate number CPHM, some position information GPS_FDC and creation-time CJSJ;

Step (3). calculate the Distance geometry mistiming of the double record of vehicle, then by Distance geometry mistiming computing velocity, and according to " grid-section numbering " mapping table, this speed is added in the section numbering at the information place, vehicle point position of second time record;

Step (4). the speed list of each vehicle on each section of this timeslice is obtained according to above step;

Step (5). identify and reject abending a little of causing due to " pseudo-stopping behavior " in speed list, and return speed list, process is as follows:

In the data structure LDID_SpeedMap that 5-1. obtains in step (4), searching loop obtains CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 and v _i+s≠ 0,0<s<=n-i, s are the number of continuous halt, then the car speed record number δ=s+1 of " pseudo-stopping behavior " process;

5-2. calculates " pseudo-stopping behavior " car speed sudden change threshold value SPSSC according to formula (2); When i ≠ 1, the formula left side is the average velocity of all speed records before this speed list first halt, as i=1, with the average velocity v in a moment on this section _tas average velocity threshold value;

$SPSSC=\left\{\begin{array}{c}\frac{\underset{f=1}{\overset{i-1}{\&Sigma;}}{v}_f}{i-1}*\mathrm{\&delta;},i\&NotEqual;1\\ v_t*\mathrm{\&delta;},i=1\end{array}\right\}---\left(2\right)$

If 5-3. is v _i+s>=SPSSC, then this speed record is the mutating speed that " pseudo-stopping behavior " causes, and rejects speed record and the mutating speed speed record of continuous halt from speed list;

5-4. completes " pseudo-stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

2. as claimed in claim 1 a kind of towards road real-time speed calculate floating car data parking behavior pattern cleaning method, it is characterized in that: described cleaning method also comprises the steps:

Step (6). identify and reject abending a little of causing due to " really stopping behavior " in speed list, and return speed list, process is as follows:

6-1. after step (5) from new LDID_SpeedMap searching loop obtain CPHM in LDID _kspeed list speedList _j, searching loop speedList _jin all speed v _iif, v _i=0 and v _i+s≠ 0,0<s<=n-i, s are the number of continuous halt; If s>=3, think that this car has occurred that " stopping behavior " skips to 6-2 and judge whether to belong to " crossing stopping behavior ";

6-2. obtains the v of this stopping behavior _icorresponding LDID and GPS_FDC _i, traversal intersectionMap, if there is this LDID in intersectionMap, illustrates that this LDID is section, crossing, enters 6-3 if do not exist; From intersectionMap, obtain crossing point (GPS_QD, GPS_ZD) that this LDID is corresponding, calculate GPS_FDC by formula (3) _iget its minimum value L with the distance at crossing, if L<L_min, then think that this " stopping behavior " belongs to " crossing stopping behavior " and do not reject, otherwise enter 6-3 and judge whether to belong to " collective stops behavior ";

L＝min(abs(GPS_FDC _i-GPS_QD),abs(GPS_FDC _i-GPS_ZD)(3)

6-3. obtains the v of this stopping behavior _icorresponding CJSJ _iand GPS_FDC _iif the speed list of other vehicles occurs that " stopping behavior " obtains the GPS_FDC of its corresponding halt under this LDID of searching loop _i'and CJSJ _i', calculate Rule of judgment according to formula (4), formula (5),

abs(GPS_FDC _i-GPS_FDC _i')＜Δs_min,i≠i'(4)

abs(CJSJ _i-CJSJ _i')＜Δt_min,i≠i'(5)

If other vehicles all meet above Rule of judgment under this LDID, " collective stops behavior " does not reject to think that this stopping behavior belonging to, otherwise this stopping behavior belonging to " really stopping behavior ", and rejects the speed record of continuous halt from speed list;

6-4. completes " really stopping behavior " identifies and returns a new LDID_SpeedMap after rejecting.

3. as claimed in claim 2 a kind of towards road real-time speed calculate floating car data parking behavior pattern cleaning method, it is characterized in that: described cleaning method also comprises the steps:

Step (7). after above step, calculate the average velocity of all speed records corresponding to LDID and be stored in database real-time speed table, update time, sheet repeated step (2) to step (6).

4. a kind of floating car data parking behavior pattern cleaning method calculated towards road real-time speed as described in one of claims 1 to 3, is characterized in that: in described step (3), velocity computation process is as follows:

3-1. obtain all records of each vehicle under this timeslice according to step (2), be designated as { CPHM _k} _ { GPS_FDC _i, CJSJ _i, k ∈ Ki ∈ n, n is the number of all records, and K is the sum of this timeslice vehicle;

3-2. calculates car speed v according to formula (1.1) or (1.2) _i, obtain car speed list { CPHM _k} _ { v _i, GPS_FDC _i, CJSJ _i.

What adopt is the absolute distance of the GPS point of double record, as the operating range of vehicle,

1) when not across using formula (1.1) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.1\right)$

2) when across using formula (1.2) when section

$v_i=\frac{abs(GPS\_{\mathrm{FDC}}_i-GPS\_{\mathrm{JD}}_i)+abs(GPS\_{\mathrm{JD}}_i-GPS\_{\mathrm{FDC}}_{i-1})}{{\mathrm{CJSJ}}_i-{\mathrm{CJSJ}}_{i-1}}---\left(1.2\right)$

GPS_JD _ifor the GPS point of double record, when across section, the some position information of the intersection point in two sections.

5. a kind of floating car data parking behavior pattern cleaning method calculated towards road real-time speed as described in one of claims 1 to 3, it is characterized in that: in described step (7), obtain the speed record of all car speed lists under LDID according to LDID_SpeedMap and calculate average velocity corresponding to LDID according to formula (6), and applying for other stored in database Real-time Road velometer.

$v_{avg}=\frac{\underset{k=1}{\overset{K}{\&Sigma;}}\underset{i=1}{\overset{n_k}{\&Sigma;}}{v}_i}{K\&CenterDot;n_k}---\left(6\right)$

K is the vehicle number under this LDID, n _kfor the speed record number of a kth car.