CN104282142B - Bus station arrangement method based on taxi GPS data - Google Patents

Abstract

The invention discloses a bus station arrangement method based on taxi GPS data. The bus station arrangement method includes the following steps: (1) obtaining getting on-and-off record sets of taxies; (2) in a territory DP with a fixed range, obtaining all getting on-and-off records of getting on of passengers in the territory; (3) clustering the getting on-and-off record sets obtained in the DP according to the getting-off positions of the passengers, obtaining territories where the passengers densely get off, and selecting the arrangement candidate territory DQ of a bus station; (4) in the arrangement candidate territory DQ of the bus station, obtaining the getting on-and-off records of getting on of the passengers; (5) as for the getting on-and-off record sets obtained in the DQ, obtaining all getting on-and-off records of the getting-off positions in the DP, wherein if the dense getting-off positions of the passengers are located in the DP, the DQ is the finally-obtained bus station arrangement territory.

Description

A kind of bus platform method to set up based on GPS data from taxi

Technical field

The present invention relates to field of computer data processing, particularly a kind of suitable for the one of the process of urban transportation division data Plant the bus platform method to set up based on GPS data from taxi.

Background technology

Since the twentieth century middle period, the growth of Global Urbanization resident population has become irresistible trend.In After reform and opening-up, urbanization process is mainly reflected in Population Urbanization and urban land Rapid Expansion for state.Chinese Government is The urban planning process planning for announcing 2014 to the year two thousand twenty new round.Urban sprawl is caused by new residential block and economic development The new-type city proper that area is constituted spreads all over all over China.With the trend of Population Urbanization, newly-built residential block resident inhabitation amount rapidly increases It is long, and have substantial amounts of resident trip daily.Crowd's trip for convenience suchly, city traffic control department need to lack to such The area of infrastructure carries out Urban Traffic Planning and builds corresponding infrastructure.

Traffic data has the feature of " big datas " such as " extensive ", " multiple types ", " high value ", " rapid growths ".It is special It is not the popularization with GPS technology, the vehicle fleet size for installing GPS is more and more, the gps data that vehicle is produced constantly increases.By Lack infrastructure in area for settlement, Residents can only select taxi as trip instrument, therefore produced by taxi The gps data value important for traffic programme has.Resident takes trip of taxi amount can reflect local resident's trip Situation, newly-built residential block resident rides, and quantity is big, and traffic pressure is also larger.In order to alleviate the traffic pressure of locality, reduce occupying People's trip expense, local traffic control department need to carry out public traffic circuit planning for these areas, set up bus platform.

In order to obtain the effective bus platform plan of establishment, industrial quarters and academia enter to path planning from different angles Numerous studies are gone.For example, Z Huang etc. are in " A Hierarchical Approach to Optimizing Bus Stop For the urban design in fast development in Distribution in Large and Fast Developing Cities " A kind of heuristic of optimization bus platform distribution, the type of bus platform include：Connection platform, crucial platform and ormal station Platform.Chao Chen etc. are in " B-Planner:Planning Bidirectional Night Bus Routes Using The extensive gps data of use is proposed in Large-Scale Taxi GPS Traces " carries out night public transport bidirectional line The problem of planning, constructs night public bus network hotspot path figure, then produces two-way public transport according to the planning of bi-directional design principle Circuit.But, in the method, bus platform is set up previously according to empirical value, for the bus station for not setting up bus platform is arranged Region selects not accounting for.In sum, obtained by the main rule of thumb doctrine of the setting of current bus platform, the rule of science are lacked Draw foundation.

The content of the invention

It is contemplated that overcoming the deficiencies in the prior art, there is provided a kind of bus station based on GPS data from taxi Platform method to set up, to support city traffic planning, can support bus station in decision-making traffic route planning process The selection of region is set.

The present invention includes five steps based on the bus platform method to set up of GPS data from taxi：

Step 1：Concentrate from GPS data from taxi and obtain taxi and get on or off the bus set of records ends BA.Taxi is at interval of fixation Time sends a GPS record information to traffic control center, and the record is included：Taxi device number, record position latitude letter Breath, record position longitude information, record moment and taxi passenger carrying status information.There are the GPS records that many taxis send Set is GPS data from taxi collection, can be expressed as TT={ tt₁,tt₂,..,tt_n}.Wherein, the number that n is recorded in representing TT Amount.

Taxi GPS records attribute tuple more than, (1≤i≤n) data record i-th in GPS data from taxi collection TT tt_iIt is expressed as tt_i=(v_i,lc_i.<llt_i,lln_i>,t_i,s_i).Wherein, v_iRepresent the device number of taxi, lc_i.<llt_i,lln_i >Represent and hire out truck position (llt_iAnd lln_iThe latitude and longitude of record position are represented respectively), t_iRepresent record moment, s_iRepresent The taxi whether status information of carrying.

Record data of getting on or off the bus set is concentrated from GPS data from taxi and is obtained, and record of getting on or off the bus per bar includes passenger once Take the record of taxi, including passenger's taxi device number, boarding position, position of getting off, get on the bus moment and the moment of getting off. The key for obtaining record data record of getting on or off the bus is to judge passenger loading position and position of getting off.

Two continuous record tt that same taxi is produced in TT_k-1And tt_k(2≤k≤n), if the carrying shape of taxi State is changed into passenger carrying status, i.e. s from zero load_k-1=0 and s_k=1, then tt_kIn record position lc_k.<llt_k,lln_k>It is exactly passenger Boarding position.Two continuous record tt that same taxi is produced in TT_lAnd tt_l+1(1≤l≤n-1), if taxi Passenger carrying status are changed into Light Condition, i.e. s from carrying_l=1 and s_l+1=0, then tt_lIn positional information lc_l.<llt_l,lln_l>Just It is the position of getting off of passenger.

The record data set of records ends of getting on or off the bus obtained from TT can be expressed as BA={ ba₁,ba₂,…,ba_m, wherein, M represents the quantity of record of getting on or off the bus.In BA, i-th record of getting on or off the bus is represented by：ba_j=(v_j,blc_j.<blt_j,bln_j>, bt_j,alc_j.<alt_j,aln_j>,at_j), j takes random natural number in 1～m.Wherein, v_jRepresent the device number of taxi, blc_j< blt_j,bln_j>(blt_jAnd bln_jBlc is represented respectively_jLatitude and longitude) represent the boarding position of passenger, bt_iRepresent on passenger Car moment, alc_j.<alt_j,aln_j>(alt_jAnd aln_jAlc is represented respectively_jLatitude and longitude) represent the position of getting off of passenger, at_jRepresent the passenger getting off car moment.

Step 2：In a region D with fixed range_PInterior, the set expression of record of getting on or off the bus is CP, the i.e. region All records of getting on or off the bus of interior passenger loading.

D_PIt is attribute tuple more than, can be specifically described as D_P=(P.<lt_P,ln_P>, a, b), wherein, P is given Particular location, lt_PRepresent the latitude of position P, ln_PThe longitude of position P is represented, a represents lt_PSearch length, b represents ln_PSearch Suo Changdu.

Get on or off the bus per bar the passenger loading positional information that contains in record, therefore, it is possible to judge that this is got on or off the bus, record is No is from D_PThe record of getting on or off the bus for setting out.If ba_j(1≤j≤m) boarding position blc_j.<blt_j,bln_j>Meet：lt_P-a≤blt_j ≤lt_P+ a and ln_P-b≤bln_j≤ln_P+ b, then ba_jIt is a record of getting on or off the bus during CP gathers.The acquisition process of CP set is needed The record of getting on or off the bus in whole BA data acquisition systems is traveled through, the final data acquisition system CP for obtaining can be expressed as CP={ cp₁,cp₂, cp₃,…,cp_cm, wherein, cm gets on or off the bus in representing CP the quantity of record.

Step 3：Record of getting on or off the bus in CP, is clustered according to passenger getting off car position, is obtained passenger getting off car position close The region of collection, and the candidate region of bus platform is set, it is expressed as D_Q。

In order to obtain the candidate region for arranging bus platform, the record of getting on or off the bus in needing to CP, according to passenger getting off car position Put and clustered, reach domain so as to find out high density, that is, position aggregation place of getting off.During calculating high density arrival domain, often Bar record of getting on or off the bus is required to calculate its all record for getting off in the nearby sphere of position, need according between geographical position away from From being judged.For geographical position l_α.<lt_α,ln_α>(lt_αAnd ln_αL is represented respectively_αLatitude and longitude) and l_β.<lt_β,ln_β >(lt_βAnd ln_βL is represented respectively_βLatitude and longitude), distance between themCan be calculated by Euclidean distance formula Arrive：

The g article record cp of CP_g(1≤g≤cm) has identical attribute with the record of getting on or off the bus in BA, can represent For：cp_g=(cv_g,cblc_g.<cblt_g,cbln_g>,cbt_g,calc_g.<calt_g,caln_g>,cat_g).Wherein, cblc_g.< cblt_g,cbln_g>And calc_g.<calt_g,caln_g>The boarding position of passenger and position of getting off, cblt are represented respectively_gAnd cbln_g Boarding position cblc is represented respectively_gLatitude and longitude, calt_gAnd caln_gThe position calc that gets off is represented respectively_gLatitude and Jing Degree, cv_gRepresent taxi device number, cbt_gRepresentative is got on the bus the moment, cat_gRepresentative is got off the moment.The process of high density domain cluster, it is main Will be using positional information of getting off.

For cp_gThe positional information calc that gets off included in (1≤g≤cm)_g.<calt_g,caln_g>, its neighborhood is expressed asWhole CP is traveled through, can be got off position calc with computed range_gRadius r In the range of all records set.

The present invention employs the concept of neighborhood during cluster, but neighborhood union can not find out certain position of getting off All neighbor points.It is probably the boundary point of neighborhood by the point of neighborhood search, getting off a little in these boundary point contiguous ranges Same clustering.Therefore, the present invention devises a 2-level region of search for the cluster process that high density reaches domain Which includes neighborhood ND_r(calc_g) in record and these record neighborhoods in all records.

In CP set, the high density that cluster, judgement are obtained reaches domain can multiple, i.e., one high density reaches domain row Table, is expressed as AD={ ad₁,ad₂,…,ad_am, wherein, am is the quantity that high density reaches domain.

For each high density that cluster is obtained reaches domain ad_s(1≤s≤am), is made up of some records of getting on or off the bus, and is represented For ad_s={ ad_s1,ad_s2,…,ad_sdm, dm represents the quantity of record of wherein getting on or off the bus.High density reaches the record ad in domain_st (1≤t≤dm) is comprising the positional information adlc that gets off_st.<adlt_st,adln_st>, adlt_stAnd adln_stPosition of getting off is represented respectively Latitude and longitude.

High density reaches ad in domain_sCenter is denoted as：Q.<lt_Q,ln_Q>, wherein, lt_QAnd ln_QCenter is represented respectively Latitude and longitude, computational methods are as follows：

According to the center Q for finding out, it may be determined that fixed range D_Q=(Q.<lt_Q,ln_Q>, a, b), wherein, a and B represents lt respectively_QAnd ln_QSearch length, D_QThe candidate region that can be arranged as bus platform.

Step 4：Obtain D_QAll set of records ends CQ of getting on or off the bus of interior passenger getting on/off.

During taking CQ set, need also exist for traveling through the record of getting on or off the bus in whole BA data acquisition systems, find on passenger Truck position is in D_QIn the range of all set of records ends of getting on or off the bus.If ba_j(1≤j≤m) boarding position blc_j.<blt_j,bln_j>It is full Foot：lt_Q-a≤blt_j≤lt_Q+ a and ln_Q-b≤bln_j≤ln_Q+ b, then ba_iIt is a record of getting on or off the bus during CQ gathers.Finally The data acquisition system CQ of acquisition can be expressed as CQ={ cq₁,cq₂,…,cq_qm, wherein, qm is the quantity that records in set.

Step 5：Record of getting on or off the bus in for CQ goal sets, acquisition get off position in D_PInterior all records of getting on or off the bus Set, is expressed as AC.If passenger getting off car position is intensive positioned at D_PIt is interior, then D_QIt is that the final bus platform for obtaining is configured to Domain.

The w article record cq of CQ_w(1≤w≤qm) has identical attribute with the record of getting on or off the bus in BA, can represent For：cq_w=(sv_w,sblc_w.<sblt_w,sbln_w>,sbt_w,salc_w.<salt_w,saln_w>,sat_w).Wherein, sblc_w.< sblt_w,sbln_w>And salc_w.<salt_w,saln_w>The boarding position of passenger and position of getting off, sblt are represented respectively_wAnd sbln_w Boarding position sblc is represented respectively_wLatitude and longitude, salt_wAnd saln_wThe position salc that gets off is represented respectively_wLatitude and Jing Degree, sv_wRepresent taxi device number, sbt_wRepresentative is got on the bus the moment, sat_wRepresentative is got off the moment.

When set is got off in calculating, need to travel through the record of getting on or off the bus in CQ, if cq_wGet off position salc_w.<salt_w, saln_w>Meet：lt_P-a≤salt_w≤lt_P+ a and ln_P-b≤saln_w≤ln_P+ b, then cq_wIt is that in set AC gets on or off the bus Record.Whole CQ record sets are traveled through, can be found out in D according to above-mentioned condition_PThe set of records ends of getting off of lower visitor, is expressed as AC={ ac₁, ac₂,…,ac_am', wherein, am ' represents the quantity of record of getting on or off the bus in gathering.

If the quantity recorded in AC is more than initial threshold dens, then D_QIt is that the final bus platform for obtaining is configured to Domain.The present invention pre-sets threshold value dens during the record quantity that carries out getting on or off the bus compares, judges, dens's Concrete setting determines that according to the natural law of gathered data collection per day threshold reference is set to 35, then the span of dens is { 35,70 ..., 35nday }, wherein, nday represents the natural law of record；

With the increase of taxis quantity, the amount of GPS data produced by taxi is increasing.In prior art, hire out The gps data collection of car, just includes record of getting on or off the bus, that is, the place position got on or off the bus and time, the GPS data from taxi of magnanimity The process of collection contributes to designing new bus platform method to set up with analysis.

The present invention is primarily based on GPS data from taxi and concentrates the record data set of getting on or off the bus for obtaining, rather than artificially builds If experience carries out bus platform setting；Then in a region D with fixed range_PIt is interior, obtain all upper and lower in the range of being somebody's turn to do Car set of records ends；Simultaneously to D_PThe record of getting on or off the bus of middle acquisition, is clustered according to passenger getting off car position, is judged passenger getting off car position Intensive region is put, and the candidate region D of bus platform is set_Q；Then obtain D_QIt is interior should in the range of all record sets of getting on or off the bus Close；The last D in acquisition_QInterior gets on or off the bus in record, and acquisition gets off position in D_PInterior all set of records ends of getting on or off the bus, if Passenger getting off car position is intensive positioned at D_PIt is interior, then D_QIt is that the final bus platform for obtaining arranges region.

Compared with prior art, effect of the invention is embodied in：

1) method to set up of bus platform is got on or off the bus data acquisition system based on taxi, in reflection crowd's Move Mode and people It is in terms of stream direction, more objective, it is credible.

2) the Density Clustering method in data mining is incorporated into into path planning problem, using density clustering algorithm pair Record of getting on or off the bus per bar is clustered, and is obtained high density and is reached domain, and then selects bus platform candidate to arrange region.

3) habit during cluster according to passenger's trip, devising a 2-level region of search can be more complete Find out set of records ends of getting on or off the bus.

4) design principle according to two-way traffic stream, judges the setting region of bus platform.

Description of the drawings

Fig. 1 flow charts of the present invention.

To be planned position views of the Fig. 2 for the embodiment of the present invention.

Fig. 3 is that passenger gets on and off position view in embodiment of the present invention data set.

Specific embodiment：

As shown in figure 1, the invention discloses a kind of bus platform method to set up based on GPS data from taxi, including with Lower step：Step 1：Concentrate from GPS data from taxi and obtain taxi and get on or off the bus set of records ends BA；Step 2：Have at one solid Determine the region D of scope_PInterior, the set expression of record of getting on or off the bus is all records of getting on or off the bus of passenger loading in CP, the i.e. region； Step 3：Record of getting on or off the bus in CP, is clustered according to passenger getting off car position, judges the intensive region in passenger getting off car position, And the candidate region of bus platform is set, it is expressed as D_Q.Step 4：In D_QInterior, set of records ends of getting on or off the bus is expressed as CQ, i.e. the region The all records of getting on or off the bus of interior passenger getting on/off；Step 5：Record of getting on or off the bus in for CQ goal sets, obtains position of getting off and exists D_PInterior all set of records ends of getting on or off the bus, if passenger getting off car position is intensive positioned at D_PIt is interior, then D_QIt is the final public transport for obtaining Platform arranges region.

Described log data set of getting on or off the bus is specifically described as BA={ ba₁,ba₂,…,ba_m, m represents record of getting on or off the bus Quantity.In BA, j-th strip is got on or off the bus and records ba_j=(v_j,blc_j.<blt_j,bln_j>,bt_j,alc_j.<alt_j,aln_j>,at_j), j takes 1 Random natural number v in～m_jRepresent the device number of taxi, blc_j.<blt_j,bln_j>(blt_jAnd bln_jLatitude and Jing are represented respectively Degree) represent the boarding position of passenger, bt_jRepresent passenger loading moment, alc_j.<alt_j,aln_j>(alt_jAnd aln_jLatitude is represented respectively Degree and longitude) represent the position of getting off of passenger, at_jRepresent the passenger getting off car moment.

The region of the fixed range that starting is selected can be described as D_P=(P.<lt_P,ln_P>, a, b), P is given concrete Position, lt_PRepresent the latitude of P points, ln_PThe longitude of P points is represented, a represents lt_PSearch length, b represents ln_PSearch length.D_P The set of records ends of getting on or off the bus of interior acquisition is expressed as CP={ cp₁,cp₂,cp₃,…,cp_cm, cm gets on or off the bus in representing CP the number of record Amount.

Getting on or off the bus in CP has recorded used in clustering according to position of getting off the concept of 2-level region of search.It is poly- Class process needs to find out the record in all neighborhoods in the range of the position Distance Judgment record radius r that gets off, and these are recorded It is added to 2-level region of search.Then the neighborhood for having added to these records in 2-level is calculated, 2- will be not added into Data record in level region of search continues to be added to region of search.

Calculate 2-level region of search during, per bar on get on or off the bus record be required to calculate its position nearby sphere of getting off Interior all records, need to calculate two geographical position l_α.<lt_α,ln_α>And l_β.<lt_β,ln_β>Between Euclidean distance Computational methods are as follows：

If the quantity recorded in the 2-level region of search that fruit is found out is more than dens, then this 2-level region of search is one High density reaches domain.

The high density found out reaches domain may multiple, be expressed as AD={ ad₁,ad₂,…,ad_am, am is that high density is arrived Up to the quantity in domain.

High density reaches domain ad_sAd can be expressed as_s={ ad_s1,ad_s2,…,ad_sdm(1≤s≤am), dm representatives are wherein The quantity of record.High density reaches the record ad in domain_st(1≤t≤dm) is comprising the positional information adlc that gets off_st.<adlt_st, adln_st>, adlt_stAnd adln_stThe latitude and longitude of position of getting off are represented respectively.High density reaches the center in domain and exactly needs Center to be found out, the latitude lt of center Q_QWith longitude ln_QComputational methods are as follows：

According to the center Q for finding out, it may be determined that fixed range D_Q=(Q.<lt_Q,ln_Q>, a, b), wherein, a and B represents lt respectively_QAnd ln_QSearch length, D_QThe candidate region that can be arranged as bus platform.

Obtain D_QInterior set of records ends CQ of getting on or off the bus, can be expressed as CQ={ cq₁,cq₂,…,cq_qm, wherein, qm is represented Get on or off the bus in CQ the quantity of record.Record of getting on or off the bus in for CQ goal sets, acquisition get off position in D_PIt is interior it is all on Get off set of records ends, be expressed as AC={ ac₁,ac₂,…,ac_am, wherein, the quantity of record of getting on or off the bus in am representative set AC. If quantity am recorded in AC is more than initial threshold dens, then D_QIt is that the final bus platform for obtaining arranges region.

Embodiment 1

The place to be planned that the present embodiment is selected is Kunming Dongcheng world Jia Shiyuan, it is thus necessary to determine that plan bus routes for which During need arrange bus platform arrange region.

Table 1. plans place in parameter information

During bus platform setting is carried out using method proposed by the present invention, need to choose a fixed position in advance P.Taxi GPS path data collection comes from Kunming 7435 and measures 7 days carrying data of hiring a car.Planning place is given in table 1 D_PParameter information.Place D to be planned_PSchematic diagram is as shown in Figure 2.

For continuous one group of taxi path data records (DEV1004,35.811024,98.541920,11:01: 10,0)～(DEV1004,35.811019,98.551020,11:35:20,0), its upper and lower visitor's record schematic diagram is as shown in Figure 3. One group is continuously recorded, when in carrying record status information when 0 is changed into 1, first state is 1 to be recorded as visitor's note Record；When carrying recording status is when 1 is changed into 0, last state be 1 being recorded as under visitor record.By upper objective and lower visitor's record Record (the DEV1004,35.811023,98.545057,11 that gets on or off the bus can be obtained:15:00,35.811019, 98.551019,11:35:15)。

Set of getting on the bus is calculated in the range of place to be planned, all record numbers of getting on or off the bus that this position is got on the bus after aggregation, are counted Amount, as a result as shown in table 2.

2 odd-numbered day of table D_PInside get on the bus record quantity of getting on or off the bus

Date	4.21	4.22	4.23	4.24	4.25	4.26	4.27
								Record quantity	327	332	302	319	321	361	329

Then to find out get on the bus set in get on or off the bus record clustered and found high density reach domain, it is involved Parameter information is as shown in table 3.

Table 3 calculates high density and reaches the parameter information that domain is related to

Parameter	Description	Value
			r	Search radius	150
dens	The density threshold for pre-setting	245

Quantity maximum 6 (Cluster0～Cluster5) in cluster is selected after cluster, record number of wherein getting on or off the bus is counted Amount, statistical result are as shown in table 4.Through judging, it is found that Cluster0 is that found out high density reaches domain.

Get on or off the bus in 4 TOP-6 clustering clusters of table record quantity

Clustering cluster	Cluster0	Cluster1	Cluster2	Cluster3	Cluster4	Cluster5
							Record quantity	426	74	31	29	27	22

In Cluster0 record according to get off position according to center computing formula be calculated Q. (24.980884, 102.764890), while according to the place D to be planned shown in Fig. 2_PSchematic diagram equally confirms D_QTerritorial scope, wherein parameter The value of a and b is identical with the value arranged in table 1.

Boarding position is calculated in D_QInterior all set of records ends of getting on or off the bus, and the odd-numbered day quantity of record of wherein getting on or off the bus is counted, Statistical result is as shown in table 5.

5 odd-numbered day of table D_QInside get on the bus record quantity of getting on or off the bus

Date	4.21	4.22	4.23	4.24	4.25	4.26	4.27
								Record quantity	601	598	602	661	645	590	589

Boarding position is in D_QAfter the completion of interior all set of records ends of getting on or off the bus are calculated, needs search out position of getting off in D_PIn All set of records ends of getting on or off the bus, then counting the wherein odd-numbered day gets on or off the bus the quantity of record, and statistical result is as shown in table 6.

6 odd-numbered day of table D_PInside get off record quantity of getting on or off the bus

Date	4.21	4.22	4.23	4.24	4.25	4.26	4.27
								Record quantity	45	39	49	56	59	43	38

In table 6, record total quantity is 329, has exceeded density threshold dens set in advance in table 3.

Therefore, by Q. (24.980884, the D for 102.764890) determining_QIt is to propose one kind based on taxi according to the present invention The target platform that the bus platform method to set up of gps data is found out is configured to domain.

The invention provides a kind of bus platform method to set up based on GPS data from taxi, implements the technical side The method and approach of case is a lot, and the above is only the preferred embodiment of the present invention, it is noted that for the art For those of ordinary skill, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvement Also protection scope of the present invention is should be regarded as with retouching.In the present embodiment, clearly each ingredient not can use prior art in addition Realize.

Claims (6)

1. a kind of bus platform method to set up based on GPS data from taxi, it is characterised in that：Comprise the following steps：

Step 1：Concentrate from GPS data from taxi and obtain cab-getter and get on or off the bus set of records ends BA；

Step 2：In a region D with fixed range_PIt is interior, it is set CP, the i.e. region by the set expression of record of getting on or off the bus The all records of getting on or off the bus of interior passenger；

Step 3：Record of getting on or off the bus in set CP, is clustered according to passenger getting off car position, judges that passenger getting off car position is close The region of collection, and select the candidate setting region D of bus platform_Q；

Step 4：By region D_QOn-board and off-board set of records ends of inside getting on the bus is expressed as CQ, i.e., in the region, cab-getter is all gets on or off the bus Record；

Step 5：Position get off in region D in obtaining set CQ_PThe set AC of interior all records of getting on or off the bus, if passenger getting off car Position is intensive positioned at D_PIt is interior, then D_QIt is that the final bus platform for obtaining arranges region.

2. the bus platform method to set up based on GPS data from taxi according to claim 1, it is characterised in that hire out Car gps data collection is the set with regard to taxi status information record, is expressed as TT={ tt₁,tt₂,..,tt_n, wherein, n generations The quantity of gps data record in table TT；

GPS data from taxi records tt_iIt is attribute tuple, tt more than one_i∈ TT, i 1～n of span, is expressed as tt_i=(v_i, lc_i.<lt_i,ln_i>,t_i,s_i), v_iRepresent the device number of taxi, lc_i.<llt_i,lln_i>Represent and hire out truck position, llt_iWith lln_iThe latitude and longitude of record position, t are represented respectively_iRepresent record moment, s_iWhether the state of carrying is believed to represent taxi Breath；

Concentrate the cab-getter for obtaining to get on or off the bus record data set from taxi path data, be expressed as BA={ ba₁, ba₂,…,ba_m, wherein, m represents the quantity of record of getting on or off the bus；

In BA, j-th strip record of getting on or off the bus is denoted as：ba_j=(v_j,blc_j.<blt_j,bln_j>,bt_j,alc_j.<alt_j,aln_j>,at_j), j Take random natural number in 1～m, v_jRepresent the device number of taxi, blc_j.<blt_j,bln_j>Represent boarding position, blt_jAnd bln_j Latitude and longitude, bt are represented respectively_jRepresent passenger getting off car moment, alc_j.<alt_j,aln_j>Representative is got off position, alt_jAnd aln_j Latitude and longitude, at are represented respectively_jRepresent the passenger getting off car moment.

3. a kind of bus platform method to set up based on GPS data from taxi according to claim 2, it is characterised in that In a region D with fixed range_PIt is interior, D_P=(P.<lt_P,ln_P>, a, b), wherein, P is given particular geographic location, lt_PRepresent the latitude of position P, ln_PThe longitude of position P is represented, a represents lt_PSearch length, b represents ln_PSearch length；

D_PAll set of records ends of getting on or off the bus of interior passenger loading are expressed as CP={ cp₁,cp₂,cp₃,…,cp_cm, wherein, cm is represented The quantity recorded in CP.

4. a kind of bus platform method to set up based on GPS data from taxi according to claim 3, it is characterised in that Record of getting on or off the bus in CP, is clustered according to passenger getting off car position, is obtained a series of clusters, records quantity and be more than density in cluster Threshold value dens is expressed as high density and reaches the intensive region of domain, i.e. passenger getting off car；

Calculate high density to reach in the calculating process of domain, record of getting on or off the bus per bar calculates its owning in the nearby sphere of position of getting off Record, that is, calculate two geographical position l_α.<lt_α,ln_α>And l_β.<lt_β,ln_β>Between Euclidean distancelt_αAnd ln_αPoint L is not represented_αLatitude and longitude, lt_βAnd ln_βL is represented respectively_βLatitude and longitude, computational methods are as follows：

${\mathrm{Dis}}_{l_{\mathrm{\&alpha;}},l_{\mathrm{\&beta;}}}=\sqrt{{({\mathrm{lt}}_{\mathrm{\&alpha;}}-{\mathrm{lt}}_{\mathrm{\&beta;}})}^2+{({\ln }_{\mathrm{\&alpha;}}-{\ln }_{\mathrm{\&beta;}})}^2},$

For cp_gIn the positional information calc that gets off that includes_g.<calt_g,caln_g>, calt_gAnd caln_gCalc is represented respectively_pLatitude Degree and longitude, 1≤g≤cm, its neighborhood are denoted as：Represent distance to get off Position calc_gThe set of all records in the range of radius r；

Get off position calc_gCluster process uses 2-level region of searchJudged,In record include neighborhood ND_r(calc_g) in record and these record neighborhoods in all records；

In CP set, cluster, judge that the high density for obtaining reaches domain for more than one, i.e., one high density reaches domain list, table It is shown as AD={ ad₁,ad₂,…,ad_am, am is the quantity that high density reaches domain；

For each high density that cluster is obtained reaches domain ad_sIt is made up of some records of getting on or off the bus, 1≤s≤am is expressed as ad_s= {ad_s1,ad_s2,…,ad_sdm, dm represents the quantity of record of wherein getting on or off the bus；

High density reaches the record ad in domain_stComprising the positional information adlc that gets off_st.<adlt_st,adln_st>, adlt_stAnd adln_st The latitude and longitude of position of getting off, 1≤t≤dm are represented respectively；

High density reaches ad in domain_sCenter is denoted as：Q.<lt_Q,ln_Q>, wherein, lt_QAnd ln_QHigh density is represented respectively reaches domain ad_sThe latitude and longitude of center, computational methods are as follows：

${\mathrm{lt}}_Q=\frac{1}{dm}\underset{t=1}{\overset{dm}{\&Sigma;}}{\mathrm{adlt}}_{st},$

${\ln }_Q=\frac{1}{dm}\underset{t=1}{\overset{dm}{\&Sigma;}}{\mathrm{adln}}_{st};$

According to the center Q for finding out, fixed range D is calculated_Q=(Q.<lt_Q,ln_Q>, a, b), wherein, a and b point Lt is not represented_QAnd ln_QSearch length, D_QAs the candidate region that bus platform is arranged.

5. a kind of bus platform method to set up based on GPS data from taxi according to claim 4, it is characterised in that In region D_QInterior, all set of records ends of getting on or off the bus of passenger are expressed as CQ={ cq₁,cq₂,…,cq_qm, wherein qm is record in set Quantity.

6. a kind of bus platform method to set up based on GPS data from taxi according to claim 5, it is characterised in that Position get off in D in obtaining set CQ_PInterior all set of records ends of getting on or off the bus, are expressed as AC={ ac₁,ac₂,…,ac_am’, am ' Represent the quantity of record of getting on or off the bus in gathering；

If the quantity recorded in AC is more than initial threshold dens, then D_QJust it is judged to that the bus platform of final acquisition is configured to Domain, wherein, the setting of dens is determined according to the natural law of gathered data collection, per day threshold reference is set to 35, then dens takes It is worth scope for { 35,70 ..., 35nday }, nday represents the natural law of record.