CN103050005B - Method and system for space and time analysis of urban road traffic states - Google Patents

Abstract

The invention relates to a method and a system for space and time analysis of urban road traffic states. On the basis of a space and time autocorrelation theory, a basic theory of mathematical statistics is applied, and according to the real-time traffic flow data of a road network, the space and time evolution of the traffic states and the generation, development and dissipation process of congestion are analyzed. The distribution features of different road states in space and time are subjected to statistical analysis by the method, the space and time rule of the evolution of the road traffic states of the urban road network is analyzed, the congestion rush hour in time and the bottleneck road in space are identified and judged, and the formulation and implementation of traffic measures are further supported.

Description

Urban road traffic state space-time analysis method and system

Technical field

The present invention relates to traffic state analysis system, particularly relate to a kind of space-time analysis method and system of urban road traffic state.

Background technology

Traffic problems are common challenges that big city faces.In order to transport solution problem, many effort have all been paid in domestic and international city, comprise the measures such as road network enlarging, crossing optimization, Intelligent Transport Systems Construction, Transportation Demand Management.But urban road network is a comprehensively Iarge-scale system for complexity, and the traffic behavior born is interactional evolutionary process between transport need and transportation supplies.Before formulation and taking traffic management measures, if traffic behavior in transportation network (unobstructed with block up) and temporal and spatial evolution thereof accurately can not be identified, then may occur that taked measure and actual traffic situation occur deviating from space-time field, not only cannot play the positive effect of traffic management measure, even likely cause the further deterioration of traffic.

In order to ensure that traffic management measure is rationally effective, the online traffic behavior that is necessary to satisfy the need carries out detailed space-time analysis, grasp the basic time and space idea that urban road network traffic behavior develops, identify the evolution Feature of jam on space-time, for the traffic congestion that takes traffic management measures timely and effectively, alleviates provides a kind of scientific and reasonable analytical approach and technology.

In the analysis and research of existing urban road traffic state, mainly analyze from Used in Dynamic Traffic Assignment theoretical prediction, reproduce based on the spacial analytical method of graph theory and traffic simulation technical modelling and angularly start with, explore the temporal-spatial evolution characteristic of traffic behavior.Though said method can portray the traffic behavior on road network eaily, but mainly there is following problem: (1) enquiry data used or behavioral parameters are always sampling sample data, by the impact of sample size and precision, there is larger gap with actual conditions; (2) theory and means used is in order to ensure availability, there is more supposed premise, there is certain limitation in practical application; (3) there is certain difficulty from time and space dimensional analysis traffic behavior Evolution, abundant not enough to the application of Traffic monitoring data.

In recent years, along with the development of Vehicle Detection technology and intelligent transportation system, a lot of big city have accumulated a large amount of arithmetic for real-time traffic flow characteristic (comprising density, speed, flow etc.), these data reflect the real-time traffic condition of urban road, and the space-time characterisation analysis that can be road traffic state provides accurately perfect data.Although spatial statistics theory and time series analysis theory are all comparatively ripe at present, can be used for also comparing shortage from the method for the time and space two dimensional analysis road traffic states simultaneously.

In sum, be necessary to develop a kind of urban road traffic state space-time analysis method and system simultaneously considering data space-time characteristic.The present invention is according to urban road traffic network real-time traffic flow data, use the space-time characterisation of statistical principle and space-time analysis method identification city road network road traffic state, the rule of effective analysis city road network road traffic state temporal-spatial evolution, as the formulation of traffic measure and the foundation of enforcement.

Summary of the invention

For above the deficiencies in the prior art, the object of this invention is to provide a kind of urban road traffic state space-time analysis method and system, it is based on space and time self-correlation theory, use mathematical statistics basic theories, according to real-time traffic flow data on road network, analyze the temporal-spatial evolution of traffic behavior and the Emergence and Development blocked up and evanishment.The distribution characteristics of road condition on space-time using this method statistic different, analyze the time and space idea that city road network road traffic condition develops, identify and judge block up peak period and bottleneck road spatially in time, supporting formulation and the enforcement of traffic measure further.

The present invention is in order to solve existing having compared with big error and the problem lacking space-time data analytical approach with actual traffic situation in existing research, and the technical scheme of employing is: a kind of urban road traffic state space-time analysis method and system.Specifically, the present invention is based on spatial autocorrelation and time self-correlation theory, the state defining a certain bar section a certain moment is a space-time object, and so space-time autocorrelation just refers to the correlationship between the same property value of all space-time object that a certain property value of space-time object (road conditions, flow etc. as certain section moment) is adjacent.From the angle of mathematical statistics, propose the index weighing this space-time autocorrelation, from autocorrelative angle research traffic flow spatio-temporal state evolutionary process.

Object of the present invention is achieved through the following technical solutions:

The space-time analysis method of urban road traffic state, the method comprises the steps:

1) from database, read the road network traffic flow space-time data needed for computational analysis, and be stored in ephemeral data table, set up space-time foundation data system;

2) data exception identification and processing rule is utilized to identify the exception that described traffic flow space-time data exists and repair;

3) from GIS map, obtain the Space Lorentz Curve matrix between road, binding time syntople, obtain the adjacent relation matrix between space-time object;

4) according to the road network traffic flow space-time data carrying out repair process in step 2, and the space-time object adjacency matrix that step 3 obtains, and space-time auto-correlation index computation model, calculate the autocorrelative general indices of space-time and local indexes;

5) multi-angle displaying and assistant analysis are carried out, to obtain urban road traffic state to the space-time auto-correlation index calculate result drawn in described step 4.

Further, from database, read traffic flow space-time data in described step 1 to comprise:

101) querying condition is arranged to parameter arranged by user individual, as from date and time, date of expiry and time, road network scope, road network scope allows user directly to select from map, or directly inputs section title or section numbering in a database;

102) connection data storehouse, is nested in input database in query statement carries out the query manipulation of data by the querying condition that user inputs, and inquiry the data obtained leaves in an interim tables of data, as the basic data of follow-up anomalous identification and process operation.

Further, the exception existed data in described step 2 identify and the concrete steps of repairing as follows:

201) data in the ephemeral data table that produces of read step 1, judge the missing data that exists in data and misdata according to the recognition rule of abnormal data, row labels of going forward side by side, as the foundation that follow-up abnormal data reparation operates;

202) according to the anomalous identification result that disorder data recognition process markup draws, the reparation rule according to abnormal data is repaired abnormal data, repairs complete data and is input to the calculating that subsequent step carries out auto-correlation index.

Further, the concrete steps obtaining the adjacent relation matrix between space-time object in described step 3 are as follows:

301) syntople between the spatial object obtaining two space-time object places, a certain bar section that what spatial object herein referred to is exactly spatially, two sections have common node then to represent they are adjacent, otherwise non-conterminous;

302) obtain two space-time object places time object between neighbouring relations, herein because traffic flow space-time data all gathers by certain time interval on section, therefore the time range of research is divided into a lot of periods by data collection interval, if two times in adjacent time interval, then show that two time objects are adjacent;

303) according to the space-time syntople between the Space Lorentz Curve obtained and time syntople determination space-time object, if two space-time object spatially with the time on all adjacent, so they are just adjacent in time-space relationship; Any two space-time object are all drawn to their syntople by above-mentioned steps, space-time adjacency matrix can be obtained.

Further, in described step 4, space-time auto-correlation index computation model specifically comprises as follows:

401) calculating of overall auto-correlation index, overall situation auto-correlation index weighs measuring of all space-time object auto-correlation degree, what investigate is the auto-correlation relation existed in studied traffic state data between space-time object, can react the auto-correlation degree of all space-time object intuitively.

402) calculating of local auto-correlation index, local auto-correlation index weighs of auto-correlation degree between all space-time object in concrete some space-time and its space-time neighborhood to measure, investigation be auto-correlation relation in concrete some space-time object and its space-time neighborhood between all space-time object on studied traffic flow attribute space-time data.

The space-time analysis system of urban road traffic state, this system comprises:

Traffic flow spatiotemporal data warehouse module, for realizing the real-time query function of traffic flow space-time data, and owing to needing data volume to be processed very large, and analytic process needs to carry out for different time sections, different local road network, requirement on flexibility is very high, and therefore described data inquiry module can according to real needs for follow-up statistical study provides basic data;

The identification of abnormal data and reparation module, for the traffic flow space-time data drawn traffic flow spatiotemporal data warehouse module, carry out identification and the reparation of abnormal data, due to reasons such as detecting device is malfunctioning, transmission line failure, inevitably there is the situation such as mistake, disappearance in the data in database, this module processes these abnormal datas, prevents the accuracy of abnormal data impact analysis result;

The computing module of space-time auto-correlation index, for receiving the traffic flow space-time data after carrying out disorder data recognition and repair process, according to the Space Lorentz Curve obtained by GIS map and time syntople determination space-time adjacency matrix, then space-time adjacency matrix is inputted the space-time auto-correlation index computation model that the present invention proposes, carry out the calculating of space-time auto-correlation index;

Result of calculation assistant analysis and display module, for the space-time auto-correlation overall situation and partial situation index obtained the computing module of space-time auto-correlation index, carry out patterned Dynamic Display, draw trend broken line graph from time dimension and space dimension angle, distribution scatter diagram, histogram and traffic behavior show thematic maps, obtain traffic state space-time Evolution.The invention has the advantages that:

The invention has the beneficial effects as follows with the traffic flow space-time data of urban road network reality as research object, based on statistical principle, propose a kind of method and system for analyzing and explore traffic flow space-time data from the autocorrelative angle of the time and space.The method and system can identify city road net traffic state transition (generation of the such as blocking up) generation on space-time, development and evanishment, add up the different distribution characteristics of road traffic state on space-time, can science, analyze the time and space idea that city road network road traffic condition develops accurately, identification and judgement block up peak period and bottleneck road spatially in time, improve rationality and the validity of traffic measure formulation and enforcement further.

Accompanying drawing explanation

Fig. 1 is road traffic state space-time analysis system primary structure frame diagram;

Fig. 2 is data identification and repair process process flow diagram;

Fig. 3 is point hour probability statistics figure of local auto-correlation index in the distribution of each interval of all space-time object in the example road network of research;

Fig. 4 is the auto-correlation scatter diagram of all space-time object in case study road network;

Fig. 5 is point hour probability statistics figure that in the example road network of research, all space-time object distribute in auto-correlation scatter diagram;

Fig. 6 is section example road network 7:10 to 7:15 in the morning period studied being positioned at auto-correlation scatter diagram first quartile.

Fig. 7 is section example road network 7:10 to 7:15 in the morning period studied being positioned at auto-correlation scatter diagram second quadrant.

Fig. 8 is section example road network 7:10 to 7:15 in the morning period studied being positioned at auto-correlation scatter diagram third quadrant.

Fig. 9 is section example road network 7:10 to 7:15 in the morning period studied being positioned at auto-correlation scatter diagram fourth quadrant.

Figure 10 is that Dong Bianmenqiao is to the trend map of foundation raft of pontoons section statistics road conditions in one day 24 hours;

Figure 11 is arranged in the local auto-correlation index one day 24 hour trend map of Dong Bianmenqiao to all space-time object in foundation raft of pontoons section;

Figure 12 is positioned at the auto-correlation scatter diagram of Dong Bianmenqiao to all space-time object in foundation raft of pontoons section.

Embodiment

A kind of urban road traffic state space-time analysis of the present invention method, the method is based on spatial autocorrelation and time self-correlation theory, the state defining a certain bar section a certain moment is a space-time object, then space-time autocorrelation just refers to the correlationship between the same property value of all space-time object that a certain property value of space-time object (road conditions, flow etc. as certain section moment) is adjacent.From the angle of mathematical statistics, propose the index weighing this space-time autocorrelation, from the evolutionary process of autocorrelative angle research traffic flow spatio-temporal state.

Method of the present invention mainly comprises following step:

Step one: read the road network traffic flow space-time data needed for computational analysis from database.This space-time data not only has different observed readings for every bar section, is not also having different observed readings in the same time for same section, has typical space-time double properties.Digital independent detailed process is as follows:

1) in order to can real-time query road network traffic flow space-time data as required, querying condition be arranged to parameter and be arranged by user individual, as from date and time, date of expiry and time, road network scope etc.Road network scope allows user directly to select from map, or directly inputs section title or section numbering in a database.

2) connection data storehouse, is nested in input database in query statement by the querying condition that user inputs and carries out the query manipulation of data.Inquiry the data obtained leaves in an interim tables of data, as the basic data of follow-up anomalous identification and process operation.

Step 2: utilize data exception identification and processing rule to carry out identification reparation to raw data.

Due to reasons such as the malfunctioning and transmission line failure of detecting device, may there is exception in the original traffic stream space-time data inquiring about acquisition in database, data exception comprises shortage of data and mistake.The existence of disappearance and misdata, have impact on the validity of data, therefore before analyzing, needs to identify the abnormal occurrence existed in these data, and use certain method to process it, in order to avoid impact analysis result.The detailed process of disorder data recognition and reparation is as follows:

1) data in the ephemeral data table that produces of read step one, judge the missing data that exists in data and misdata according to the recognition rule of abnormal data, row labels of going forward side by side, as the foundation that follow-up abnormal data reparation operates.

2) according to the anomalous identification result that disorder data recognition process markup draws, the reparation rule according to abnormal data is repaired abnormal data, repairs complete data and can be input to the calculating that subsequent step carries out auto-correlation index.

Step 3: obtain the Space Lorentz Curve matrix between road from GIS map, in conjunction with time syntople in database, calculates the adjacent relation matrix obtained between space-time object.

According to the connotation of space-time auto-correlation index, definition and all space-time object on a space-time object adjacent space position, adjacent time point are the space-time neighborhood of this space-time object, then space-time syntople just can be obtained by Space Lorentz Curve and time syntople.Therefore, the step calculating the space-time syntople of two space-time object is as follows:

1) syntople between the spatial object obtaining two space-time object places, a certain bar section that what spatial object herein referred to is exactly spatially, two sections have common node then to represent they are adjacent, otherwise non-conterminous.

2) obtain two space-time object places time object between neighbouring relations, herein because traffic flow space-time data all gathers by certain time interval on section, therefore the time range of research can be divided into a lot of periods by data collection interval, if two times in adjacent time interval, then show that two time objects are adjacent.

3) according to the space-time syntople between the Space Lorentz Curve obtained and time syntople determination space-time object, if two space-time object spatially with the time on all adjacent, so they are just adjacent in time-space relationship.Any two space-time object are all drawn to their syntople by above-mentioned steps, space-time adjacency matrix can be obtained.

Step 4: according to the road network traffic flow space-time data repaired in step 2, and the space-time object adjacency matrix that step 3 obtains, according to the space-time auto-correlation index computation model that the present invention proposes, calculate the autocorrelative general indices of space-time and local indexes.Computation model detailed process is as follows:

1) calculating of overall auto-correlation index.Overall situation auto-correlation index weighs measuring of all space-time object auto-correlation degree, investigation be the auto-correlation relation existed in studied traffic state data between space-time object, the auto-correlation degree of all space-time object can be reacted intuitively.

2) calculating of local auto-correlation index.Local auto-correlation index weighs of auto-correlation degree between all space-time object in concrete some space-time and its space-time neighborhood to measure, investigation be auto-correlation relation in concrete some space-time object and its space-time neighborhood between all space-time object on studied traffic flow attribute space-time data.

Step 5: the multi-angle of result of calculation is shown and assistant analysis.

For the auto-correlation index calculated, general indices is a numerical value, but local auto-correlation index has different values for different space-time object, and therefore, calculation result data amount is larger.Show that these data just can be convenient to research and analyse by certain clue and logic, the means of displaying comprise a point trend broken line graph for time peacekeeping space dimension, distribution scatter diagram, histogram and traffic behavior show thematic maps.Additional transport management and researchist can analyze and study the distribution characteristics of traffic behavior on space-time and Changing Pattern.

Above-mentioned steps two, three, four is cores of this method, is described in further detail below to the computing method in step 2, three, four and process.

The concrete grammar of described step 2 to the identifying and modifying of abnormal data is as follows:

For the identification of missing data, because traffic flow space-time data is gathered by certain time interval by detecting device, so data are easy to verify.Mistiming between two adjacent datas collected is exactly 1 data collection interval under normal circumstances, more than representing to there is shortage of data between them when 1 time interval.

For the identification of misdata, there are following two kinds of methods:

1) recognition methods of conventional misdata is threshold method, exceedes certain threshold value and is just considered to misdata.

2) carry out by several traffic parameter data the method that associating judges.For a certain traffic data, according to the speed in traffic flow theory, relation between density and flow, can verify whether it is misdata by other two kinds of data.

After identifying abnormal data, just can repair it.The method of repairing has:

1) adopt the historical data of the previous day to repair, be applicable to obliterated data;

2) adopt historical trend data and measured data weighted estimation to repair, be applicable to misdata;

3) adjacent time interval data are utilized to carry out linear interpolation to repair;

4) predicted value obtaining loss period by the data of a front n period is repaired;

5) predicted value obtained by adjacent segments is repaired.

The method wherein utilizing adjacent time interval data to carry out linear interpolation to repair is concisely easy, and effect is better.Therefore be the recommend method in the present invention.For two data y(i) and y(j) (i<j), on the left of it, right side and middle missing data y(n) (i<n<j) can carry out interpolation with following formula:

$y\left(n\right)=\frac{n-i}{j-i}(y\left(j\right)-y\left(i\right))$

The concrete grammar that the space-time object adjacent relation matrix of described step 3 calculates is as follows:

According to definition, be the space-time neighborhood of this space-time object with all space-time object on a space-time object adjacent space position, adjacent time point.So, space-time syntople just can be determined by Space Lorentz Curve and sequential syntople.Because adjacent weight is a value of getting 0 or 1, get 1 expression and adjoin, get 0 expression and do not adjoin, therefore space-time neighboring rights weight values can be expressed as the product of Spatial Adjacency weighted value and time neighboring rights weight values.If space-time object ST(p, i) and ST(q, j) be section p respectively in the state in i moment and the section q state in the j moment, so space-time object ST(p, i) and ST(q, j) between adjacent weight be: w _{(p, i) (q, j)}=w _{(p, q)}× w _{(i, j)}

Wherein w _{(p, q)}spatial Adjacency weight between section p and section q:

W _{(i, j)}for the time between moment i and moment j adjoins weight:

Should be noted that, specify the space-time neighborhood each other of same moment of adjacent space position herein, but not space-time object each other between the adjacent moment of same locus, this is that this time autocorrelation may cover special heterogeneity because the state of the adjacent moment of the same space geographic object (as a certain bar section) is a time series.Therefore, in computing formula, specify same section (p=q) not spatial neighborhood each other, but synchronization (i=j) time neighborhood each other.

Calculate all w _{(p, i) (q, j)}, obtain a space-time object adjacency matrix.This step calculates the space-time object adjacency matrix obtained, the traffic flow space-time data after repairing with previous step, inputs next step together and carries out space-time auto-correlation index calculate.

Calculating concrete grammar and the process of the space-time auto-correlation overall situation and partial situation index of described step 4 are as follows:

If y _{(p, i)}space-time object ST(p, i) a certain property value (as the volume of traffic, speed, density etc.) of (state in p section i moment), so in certain hour section and spatial dimension, all space-time object can be expressed as about the overall auto-correlation index A of this property value:

$A=\frac{\mathrm{NT}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}(y_{(p,i)}-\stackrel{\_}{y})(y_{(q,j)}-\stackrel{\_}{y})}{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\_}{y})}^2\&times;{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

Wherein N be research spatial dimension in the number in all sections, T be research time range in the time interval number of data acquisition.NT is the number of all space-time object.W _{(p, i) (q, j)}for space-time object ST(p, i) and ST(q, j) between neighboring rights weight values, y is the average of all space-time object y property values:

$\stackrel{\_}{y}=\frac{1}{\mathrm{NT}}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{y}_{(p,i)}$

The value of overall situation auto-correlation index A, in-1 to 1 scope, is in the implication that different sub-ranges represents different:

1) A is greater than 0 expression positive correlation, more more obvious close to 1 this trend.Be embodied in studied property value to reach unanimity in adjacent segments adjacent moment.For speed, if the velocity amplitude of a certain section a certain period is high, the velocity amplitude of its adjacent segments adjacent time interval also trends towards high level; Velocity amplitude is low, and the velocity amplitude of its adjacent segments adjacent time interval also trends towards low value.The traffic behavior characteristic that the value of overall situation auto-correlation index A reflects is different along with studied attribute difference.If the attribute of research is speed, speed higher expression road is more unimpeded, and time so A is greater than 0, this positive correlation represents that the road network of research shows as on space-time and blocks up or the aggregation properties of unimpeded state, and more more remarkable close to 1 aggregation properties.Particularly, the even a certain section a certain period is unimpeded, and its adjacent segments is also tending towards unimpeded at adjacent time interval; Block up, its adjacent segments is also tending towards blocking up at adjacent time interval.

2) to equal 0 expression uncorrelated for A;

3) A is less than 0 expression negative correlation, and more close-1 this trend is more obvious.Be embodied in the value of studied property value in adjacent segments adjacent moment and be tending towards contrary.For speed, if the velocity amplitude of a certain section a certain period is high, then the speed of its adjacent segments adjacent time interval trends towards low value; Velocity amplitude is low, and the speed of its adjacent segments adjacent time interval trends towards high level on the contrary.If the attribute of research is speed, speed higher expression road is more unimpeded, time so A is less than 0, this negative correlation represents that in traffic behavior characteristic the road network of research shows as on space-time and blocks up or the heterogeneity of unimpeded state, and more close-1 heterogeneous character is more remarkable.Particularly, the even a certain section a certain period is unimpeded, and its adjacent segments is tending towards blocking up at adjacent time interval on the contrary; Block up, its adjacent segments is tending towards unimpeded on the contrary at adjacent time interval.

Overall situation auto-correlation index A is the index reflecting overall space-time object auto-correlation situation, cannot weigh the autocorrelative local characteristics of space-time.For concrete some space-time object, weigh the correlationship in it and its space-time neighborhood between all space-time object by another index, i.e. space-time auto-correlation local indexes:

a _(p，i)=Z _(p，i)Wz _(p，i)

Wherein: ${\mathrm{Wz}}_{(p,i)}=\frac{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}{Z}_{(q,j)}}{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

$Z_{(p,i)}=\frac{(y_{(p,i)}-\stackrel{\_}{y})}{\mathrm{\&sigma;}}$

$\mathrm{\&sigma;}=\sqrt{\frac{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\_}{y})}^2}{\mathrm{NT}-1}}$

Although a _{(p, i)}unlike auto-correlation general indices A, have fixing span, but its positive and negative situation also represent different implications:

1) a _{(p, i)}get on the occasion of, represent space-time object ST(p, i) (state in p section i moment) show as positive correlation with all space-time object in its space-time neighborhood, the larger this trend of absolute value is more obvious.For speed, if p section i moment velocity amplitude is high, the velocity amplitude of its adjacent segments adjacent time interval is also high; Velocity amplitude is low, and the velocity amplitude of its adjacent segments adjacent time interval is also low.Similar with overall auto-correlation index, if the attribute of research is speed, speed higher expression road is more unimpeded, a _{(p, i)}get on the occasion of time, ST(p, i) show as in traffic behavior characteristic with the positive correlation of all space-time object in its space-time neighborhood: if p section is unimpeded in the i moment, its adjacent segments is also unimpeded in adjacent moment; Its adjacent segments of the words of blocking up also blocks up in adjacent moment.

2) a _{(p, i)}get 0 and show space-time object ST(p, i) all space-time object are uncorrelated in (state in p section i moment) and its space-time neighborhood.

3) a _{(p, i)}get negative value, show space-time object ST(p, i) all space-time object show as negative correlativing relation in (state in p section i moment) and its space-time neighborhood, and the larger this trend of absolute value is more obvious.For speed, if p section i moment velocity amplitude is high, then the velocity amplitude of its adjacent segments adjacent time interval is low; Velocity amplitude is low, and the velocity amplitude of its adjacent segments adjacent time interval is high on the contrary.Similar with overall auto-correlation index, if the attribute of research is speed, speed higher expression road is more unimpeded, a _{(p, i)}when getting negative value, ST(p, i) show as in traffic behavior characteristic with the negative correlativing relation of all space-time object in its space-time neighborhood: if p section is unimpeded in the i moment, its adjacent segments blocks up in adjacent moment; Its adjacent segments of the words of blocking up is unimpeded in adjacent moment.

Z in formula _{(p, i)}represent space-time object ST(p, i) (state in p section i moment) standardization y property value, Wz _{(p, i)}represent space-time object ST(p, i) space-time neighborhood in the weighting standard y property value of all space-time object.Therefore, for space-time object ST(p, i) (state in p section i moment), can its Z of integrated use _{(p, i)}and Wz _{(p, i)}value weighs its relation on y property value and around it between space-time object.With Z _{(p, i)}for horizontal ordinate, Wz _{(p, i)}for ordinate draws scatter diagram, be called space-time auto-correlation scatter diagram.

In space-time auto-correlation scatter diagram, each space-time object can a corresponding point, it is positioned at different quadrants and represents space-time object and be in different conditions in space-time auto-correlation scatter diagram, therefore, can judge the distribution character of traffic behavior on space-time with space-time auto-correlation scatter diagram:

1) point being positioned at first quartile represents this space-time object ST(p corresponding to point, i) when (state in p section i moment) gets high level on studied attribute (flow, speed, density etc.), its adjacent space-time object also trends towards getting high level, shows positive correlation.Traffic behavior characteristic representated by the position of space-time object in auto-correlation scatter diagram is different along with studied attribute difference.If the attribute of research is speed, so space-time object ST(p, i) corresponding point is positioned at first quartile in traffic behavior characteristic, just represents that p section is in i moment unimpeded (speed high level), in adjacent moment also unimpeded (speed high level), p road section traffic volume state space-time feature shows as unimpeded gathering to its adjacent segments.

2) point being positioned at the second quadrant represents this space-time object ST(p corresponding to point, i) when (state in p section i moment) gets low value on studied attribute (flow, speed, density etc.), its adjacent space-time object trends towards getting high level, shows negative correlativing relation.If the attribute of research is speed, so space-time object ST(p, i) corresponding point is positioned at the second quadrant in traffic behavior characteristic, just represents that p section blocks up (speed low value) in the i moment, but its adjacent segments is in adjacent moment unimpeded (speed high level), p road section traffic volume state space-time feature show as take the lead in blocking up, delayed dissipation.

3) point being positioned at third quadrant represents this space-time object ST(p corresponding to point, i) when (state in p section i moment) gets low value on studied attribute (flow, speed, density etc.), its adjacent space-time object also trends towards getting low value, shows positive correlation.If the attribute of research is speed, so space-time object ST(p, i) corresponding point is positioned at third quadrant in traffic behavior characteristic, just represents that p section blocks up (speed low value) in the i moment, its adjacent segments also blocks up in adjacent moment, and p road section traffic volume state space-time feature shows as gathering of blocking up.

4) point being positioned at fourth quadrant represents this space-time object ST(p corresponding to point, i) when (state in p section i moment) gets high level on studied attribute (flow, speed, density etc.), its adjacent space-time object trends towards getting low value, shows negative correlativing relation.If the attribute of research is speed, so space-time object ST(p, i) corresponding point is positioned at fourth quadrant in traffic behavior characteristic, just represents that p section is in i moment unimpeded (speed high level), but its adjacent segments blocks up (speed low value) in adjacent moment, p road section traffic volume state space-time feature shows as delayedly to block up, takes the lead in dissipating.

The invention still further relates to a kind of urban road traffic state space-time analysis system.This system comprise traffic flow spatiotemporal data warehouse module, abnormal data identification and repair module, the computing module of space-time auto-correlation index and result of calculation assistant analysis and display module, system framework is as shown in Figure 1.

Traffic flow spatiotemporal data warehouse module: this module realizes the real-time query function to needing the traffic flow space-time data analyzed.Owing to needing traffic space-time data amount to be processed very large, and analytic process needs to carry out for different time sections, different road network scope, requirement on flexibility is very high, and therefore real-time data inquiry module can according to real needs for follow-up statistical study provides basic data.

In view of the data search efficiency of the brilliance that oracle database demonstrates when data volume is large, recommendation Oracle of the present invention is as database engine, and the System.Data.OracleClient class provided in Visual studio can realize oracle database very easily and connect and operation.In order to can data query according to demand neatly, the section set etc. that querying condition comprises as from date, initial time, date of expiry, termination time, road network is nested in query statement as variable, inputted according to demand by user during inquiry, the road network scope of research can be selected in road network map by user.

The identification of abnormal data and reparation module: this module, to the data exception existed in the traffic flow space-time data of traffic flow spatiotemporal data warehouse module gained, carries out identifying and repairing.It comprises disorder data recognition submodule and abnormal data repairs submodule two parts, is first identified the loss existed in data or mistake, then transfer to abnormality processing submodule to process the result of identification by anomalous identification submodule.Disorder data recognition and reparation flow process are as shown in Figure 2.

Disorder data recognition flow process comprises following step:

1) from i 1, obtain the time series data of the i-th a certain attribute of the traffic flow of section bar within search time scope (flow, speed etc.), arrange according to acquisition time sequencing.

2) according to research time period in total data collection interval number N initialization one be entirely 0 array S [i].

3) travel through the time series data in this section, if outside certain threshold value, be expressed as abnormal data, do not carry out operating just forwarding next data to; Otherwise, put into array S [i] relevant position by the time period residing for each data, forward next data to.Repetitive operation is until traveled through the time series in this section.Like this, be that the position of 0 lacks or misdata exactly in array S [i], non-vanishing data are exactly normal data.Store this section anomalous identification result array S [i], so that carry out abnormal reparation according to it.

4) above-mentioned steps is repeated, until all sections have all traveled through.

Dealing of abnormal data flow process comprises following step:

1) from i=1, the anomalous identification result array S [i] in i-th section is obtained.

2) first of calculating in array S [i] is not the index a of the number of 0, and last is not the index b of the number of 0.

3) travel through all items in array in S [i], if not 0, represent normal data, do not carry out operating just forwarding array the next item down to; Otherwise, represent shortage of data or mistake, the item of index before a just carries out extrapolation reparation by the normal data after a, and the item of index between a, b just does interpolation reparation by its adjacent normal data, and the item of index after b just carries out extrapolation reparation by the normal data before b.Repetitive operation is until the element in array S [i] has been traversed, and the Time-space serial data in this section are just repaired complete.

4) above-mentioned steps is repeated, until all sections are all repaired complete.

The computing module of space-time auto-correlation index: this module is made up of space-time adjacency matrix calculating sub module and auto-correlation index calculate submodule two parts.This module receives the identification of abnormal data and repairs the complete traffic flow space-time data of resume module, first the Space Lorentz Curve obtained according to GIS map and time syntople determination space-time adjacency matrix, then space-time adjacency matrix is substituted into the space-time auto-correlation index computation model proposed, carry out the calculating of space-time auto-correlation index.According to method of the present invention, result of calculation comprises overall space-time auto-correlation index A, the local auto-correlation index a of each space-time object _{(p, i)}, standardized nature value Z _{(p, i)}with the weighting standard property value Wz of all space-time object in the space-time neighborhood of space-time object _{(p, i)}.

Result of calculation assistant analysis and display module; The space-time auto-correlation overall situation and partial situation index that this module obtains for the computing module analyzing space-time auto-correlation index, carry out graphical representation, divide time peacekeeping space dimension to draw trend broken line graph, distribution scatter diagram, histogram and traffic behavior show thematic maps, analyze traffic flow spatio-temporal state evolutionary process.GDI in Visual studio provides a series of drawing instrument, can be used for automatically drawing in the present invention the statistical graph needing to show.Meanwhile, GIS secondary development bag provides the tool box drawing various thematic maps, can be shown the automatic drafting of thematic maps by programming realization traffic behavior.

Specifically, this module can to each section, by the local auto-correlation index a of all space-time object on this section _{(p, i)}draw broken line graph in chronological order, study this section and the correlativity in section around it; The standardized nature value Z of space-time object in whole road network, a certain bar section or several sections will be belonged to _{(p, i)}with the weighting standard property value Wz of all space-time object in its space-time neighborhood _{(p, i)}be depicted as space-time auto-correlation scatter diagram, the traffic flow modes characteristic of entirety or local road network can be studied; In a certain data acquisition intervals, each road only has data, therefore a space-time object is also only had, can according to the distribution of these roads all quadrants in space-time auto-correlation scatter diagram (as previously mentioned, the road being in different quadrant can present different qualities), difference carried out to the section of different qualities painted, be depicted as a traffic behavior and show thematic maps.Traffic behavior is drawn to each period and shows that thematic maps carries out dynamic demonstration, can the figuratively spatial and temporal distributions of bright road traffic state and evolution properties.

Above-mentioned a few part of module is that the direction of carrying out data process&analysis by system connects each other, and each module processes according to the data of certain operation order to input.

First, the Time and place scope of the needs inquiry that traffic flow spatiotemporal data warehouse module inputs according to user, inquires the traffic flow space-time data for analyzing from Oracle data.

Secondly, the situation of disappearance and mistake may be there is owing to inquiring about the data obtained, therefore by abnormal data identification and repair after module accepts and carry out anomalous identification and reparation, in order to avoid have an impact to analysis result.

Then, will repair complete traffic flow space-time data, the computing module being delivered to space-time auto-correlation index carries out the calculating of space-time auto-correlation index.The computing module of space-time auto-correlation index is made up of space-time adjacency matrix calculating sub module and auto-correlation index calculate submodule two parts.Space-time adjacency matrix calculating sub module first calculates the space-time adjacency matrix between space-time object, and the space-time object adjacency matrix calculated is input to auto-correlation index calculate submodule together with the complete space-time traffic flow data of reparation again and calculates the autocorrelative general indices of space-time and local indexes.

Finally, calculate the space-time autocorrelative overall situation and partial situation index of gained, be further analyzed and graphical representation by result of calculation assistant analysis and display module.

This system uses C# programming language to realize in conjunction with oracle database, GIS secondary development bag.Because data volume is large, using oracle database to carry out data storage and retrieval efficiency can be higher.Based on GIS secondary development can automatic acquisition user choose road network scope, calculate Space Lorentz Curve and carry out traffic behavior and show the drafting of thematic maps.GDI in C# provides a series of drawing instrument, is convenient to realize automatically drawing the required various statistical graphs shown in the present invention.

In order to the implementation process of urban road traffic state space-time analysis method and system is described more intuitively, the present invention is shown by an application example.

Choose the through street within five rings, Beijing, major trunk roads road network as analytic target, traffic data selects the road condition data in section.Road condition data is the index weighing the coast is clear degree, and the ratio being numerically equal to the real-time speed of a motor vehicle and design speed on section is multiplied by 100.If the speed of a motor vehicle is greater than design speed in real time, road conditions value gets 100.Therefore, road conditions value is the number of 0 to 100, and it is more more unimpeded close to 100 roads, more more blocks up close to 0 road.

Search time scope gets 1 day 24h, and the acquisition time of road condition data is spaced apart 5 minutes, therefore within one day 24 hours, there are 288 data in every bar section.Because the road condition data studying separately some day can be subject to the impact of a lot of accidentalia, and traffic data has periodicity clearly, short period is one day, long period is one week, therefore by half a year on July 1,1 day to 2010 January in 2010, the road conditions on all Mondaies on average become the data analysis of one day 288 period, to explore the universal law that Beijing's road network develops at the road traffic state on Monday.

First, carry out the inquiry of traffic flow space-time data, commencement date and deadline and time, road network scope are input in program, inquire the road conditions space-time data of the road network scope of research.

Secondly, identification and the reparation of abnormal data is carried out.Anomalous identification and reparation automatically complete on backstage after query script terminates.

Then, the calculating of space-time auto-correlation index is carried out.

The first step first obtains Space Lorentz Curve between section according to GIS map.From GIS attribute database, specifically inquire the start node in every bar section, what have common node with certain section is exactly its adjacent segments.In sequential, the time range of research is divided into some periods by data acquisition intervals and is numbered with sequence number, data acquisition intervals is 5 minutes herein, and so one day 24h just has 288 periods.The road condition data dropping on a certain period belongs to this period, represents adjacent between the period of two sequence number differences 1.Computing formula according to space-time adjacency matrix can draw the space-time adjacency matrix of all space-time object.

Second step, calculates the space-time auto-correlation overall situation and local indexes according to the computing formula in method of the present invention.Concrete calculation procedure and formula are shown in embodiment 2.

Finally, the space-time auto-correlation the calculated overall situation and local indexes are depicted as figure to show.Comprise local auto-correlation index a _{(p, i)}draw broken line graph in chronological order; From overall or draw, by Z with spatial dimension at times _{(p, i)}as horizontal ordinate, Wz _{(p, i)}for ordinate is depicted as space-time auto-correlation scatter diagram; Counting space-time object in space-time auto-correlation scatter diagram at times in the distribution frequency of all quadrants draws histogram, can study the characteristic of traffic flow modes at day part; Space-time object in the distribution of all quadrants, represents the traffic behavior residing for section in space-time auto-correlation scatter diagram, shows thematic maps according to this distribution core traffic behavior.

Through the computing of each module of system, obtain following result:

1) overall auto-correlation index

Reject the section of disappearance detector data, in 288 periods of one day 24h, there are 118368 space-time object in 411 sections on the road network studied, their overall space-time auto-correlation index is: 0.352, shows weak positive correlation.If it is high in certain period road conditions value to be embodied as certain section, the road conditions value of its adjacent segments adjacent time interval is also faintly tending towards high level; Road conditions value is low, and the road conditions value of its adjacent segments adjacent time interval is also faintly tending towards low value.From the angle of traffic behavior, the even a certain section a certain period is unimpeded, and its adjacent segments is also tending towards unimpeded at adjacent time interval; Block up, its adjacent segments is also tending towards blocking up at adjacent time interval.Therefore, the aggregate performance on space-time of whole road network is block up or the aggregation properties of unimpeded state.

2) local auto-correlation index

Divide a hour local auto-correlation index a for statistics space-time object _{(p, i)}in the distribution situation of each numerical intervals, be depicted as histogram, as shown in Figure 3.

In figure 3, local auto-correlation index a _{(p, i)}the ratio of high level (being greater than 1 part) have obvious three peak values, represent respectively the road conditions in morning high high assemble the period, with early, road conditions value low gathering period of evening peak, this show road network block up on a large scale or unimpeded time, present larger positive correlation between the road conditions value of whole road network; And entering in morning peak, morning peak dissipates and evening peak dissipates process, local auto-correlation index a _{(p, i)}the ratio of negative value (being less than 0 part) all relatively large, this explanation is blocked up with unimpeded transfer process at the whole network, increases between the road conditions value of whole road network in negative correlation.

As previously described, positive correlation represents that road network presents on space-time and blocks up or the aggregation properties of unimpeded state, namely certain section is unimpeded in a certain period, and its adjacent segments is also tending towards unimpeded at adjacent time interval, and its adjacent segments of the words of blocking up also blocks up at adjacent time interval; And negative correlation represents that road network presents the heterogeneity of blocking up with unimpeded on space-time, namely certain section is at a time unimpeded, and its adjacent segments blocks up on the contrary in adjacent moment, and its adjacent segments of the words of blocking up is unimpeded on the contrary at adjacent time interval.Therefore, when road network blocks up on a large scale, blocking up that space-time shows strengthens with unimpeded aggregation properties; And blocking up with unimpeded transfer process at road network, the aggregation that space-time shows weakens, heterogeneous enhancing.

Occur that above-mentioned phenomenon is because almost every bar road all can block up in peak period, all can have unimpeded in morning, within these periods, therefore show as blocking up or unimpeded aggregation properties on space-time more; But it is different that each bar road enters the time of blocking up, thus the time of dissipation of blocking up is also different, is therefore blocking up and is having very large difference between road conditions value in unimpeded transfer process, space-time presents and blocks up and unimpeded heterogeneity.But for some space-time object (state in a certain section a certain moment), be embodied in aggregation that unimpeded aggregation still blocks up, be unimpeded heterogeneity of being surrounded by blocking up or block up by the heterogeneity of unimpeded encirclement, differentiate from microcosmic auto-correlation scatter diagram with regard to needs.

3) auto-correlation scatter diagram

By the Z of 118368 space-time object _{(p, i)}as horizontal ordinate, Wz _{(p, i)}space-time auto-correlation scatter diagram is depicted as ordinate.As shown in Figure 4.The distributing position of 118368 space-time object in auto-correlation scatter diagram is illustrated in Fig. 4.Major part space-time object is arranged in one, three quadrants of space-time auto-correlation scatter diagram, and be positioned at two, four-quadrant point only accounts for a very little part, this also illustrates correlationship in studied road network and time range between space-time object mainly based on positive correlation, whole road network main manifestations on space-time is block up or the aggregation properties of unimpeded state.

For convenience of explanation, with hour for base unit, point the space-time object being arranged in the different quadrant of space-time auto-correlation scatter diagram to be added up in one day 24 hours, as shown in Figure 5.

As can see from Figure 5, at the space-time object (state in certain section moment) of first quartile, its road conditions show as high level and are surrounded by high level, represent unimpeded gathering, it all accounts for the principal status of public economy at 23:00 to 6:00 in morning, illustrates that during this period of time whole road network is in unimpeded state.The clustering phenomena that blocks up of three quadrant representatives, two peak values are had: morning, 7:00 to 11:00 was in four hours in one day, space-time object within per hour in three quadrants all account for more than 50%, and at 8:00 to 9:00, the space-time object in three quadrants even account for more than 70%; In afternoon 14:00 to 19:00 five hours, the space-time object within per hour in three quadrants account for more than 50%, but is all less than 60%.Result shows that evening peak is longer than the morning peak duration, and intensity of blocking up relatively is disperseed; Morning peak is more blocked up than evening peak, and particularly at 8:00 to 9:00, whole road network is in the state of blocking up most in a day.This hour of 12:00 to 13:00 at noon, the space-time object in road network there will be unexpected growth in the distribution of first quartile, has exceeded 40%, has reduced rapidly again afterwards, illustrates at this moment, and of short duration unimpeded state appears in road network.If weigh unimpeded degree with the space-time object proportion of a quadrant, then noon of short duration unimpeded after, whole road network is until just there will be the unimpeded state of same degree after 21:00.

4) traffic behavior shows thematic maps

Due in some periods, each road only has a space-time object, for ease of explanation, can according to the distribution of these roads all quadrants in space-time auto-correlation scatter diagram (as previously mentioned, the road being in different quadrant can present different qualities), difference carried out to the section of different qualities painted, be depicted as thematic maps.The thematic maps of Different periods is demonstrated continuously, vivo can show the temporal-spatial evolution characteristics of road traffic state.The traffic behavior that Fig. 6 to Fig. 9 illustrates when 7:10 to 7:15 shows thematic maps.

In figure 6, as if level bridge is now in first quartile in auto-correlation scatter diagram to sections such as wolf Mount Tai Xi Qiao, the road conditions value of road conditions value and its adjacent segments adjacent time interval is all tending towards high level, is the unimpeded gathering section of road network;

In the figure 7, Ma Jialou bridge is now in the second quadrant in auto-correlation scatter diagram to sections such as the western red raft of pontoons, and its road conditions get low value, but the road conditions value of its adjacent segments adjacent time interval is tending towards high level, and therefore this section is the congested link in road network unobstructed area.

In fig. 8, as if level bridge is now in third quadrant in auto-correlation scatter diagram to sections such as Yue Ge Zhuan Qiao, the road conditions value of road conditions value and its adjacent segments adjacent time interval is all tending towards low value, is that blocking up in road network assembles section;

In fig .9, as if level bridge is now in fourth quadrant in auto-correlation scatter diagram to sections such as yamen's mouth bridges, its road conditions get high level, but the road conditions of its adjacent segments adjacent time interval are tending towards low value, and therefore this section is the unimpeded section in road network congestion regions.

5) the auto-correlation local indexes in concrete section and auto-correlation Discrete point analysis

For ease of studying the road traffic state Changing Pattern in concrete a certain section further, in the road network of research, choose Dong Bianmenqiao to foundation raft of pontoons section as research object.Figure 10 illustrates the statistics road conditions trend on all Mondaies in this section half a year on July 1,1 day to 2010 January in 2010.Figure 11 is the auto-correlation local indexes a in this section _{(p, i)}variation diagram in time.Figure 12 is the auto-correlation scatter diagram that on this section, 288 space-time object are drawn.

As can be seen from Figure 10, there was the evening peak of the morning peak of three period: 8:00 to 11:00 that block up, the noon peak of 14:00 to 16:00 and about 18:00 in this section in one day.As can be seen from Figure 11, four periods of about the 4:00 in a day, about 9:30, about 15:00 and about 18:00, the auto-correlation local indexes in this section all get on the occasion of, and absolute value is larger.This four periods respectively in corresponding Figure 10 this section road conditions trend map on the clear area in morning, block up district and the afternoon in morning, two districts that block up at dusk.Show that this section is unimpeded or when blocking up, and has larger positive correlation with section adjacent time interval around it.And in the process of blocking up formation and dissipation, as when morning, 6:00 to 7:00 morning peak started to be formed, auto-correlation local indexes value is all lower, even there is negative value, show blocking up and the transition period of unimpeded state, this section and around it section adjacent moment positive correlation reduce, even there is negative correlativing relation.

As can be seen from Figure 12, the point that the space-time object in this section is arranged in first quartile represents the unimpeded period of Figure 10 from night to morning, and major part is distributed in straight line Wz _{(p, i)}=Z _{(p, i)}upper left side, illustrates that this section is when road network is unimpeded, and its road conditions value all will lower than the road conditions value of its adjacent road; The point that the space-time object in this section is arranged in third quadrant represents three of Figure 10 and blocks up the period.In this section, unimpeded and between blocking up the space-time object of transition period be all positioned at the second quadrant, when illustrating that this section blocks up in this transient process, its adjacent segments is unimpeded, therefore this section shows as and takes the lead in blocking up when blocking up and occurring in traffic behavior characteristic, to block up delayed dissipation when dissipating, in road network, belonging to local block up and cause section or Bottle Neck Road.

Should be appreciated that above is illustrative and not restrictive by preferred embodiment to the detailed description that technical scheme of the present invention is carried out.Those of ordinary skill in the art can modify to the technical scheme described in each embodiment on the basis of reading instructions of the present invention, or carries out equivalent replacement to wherein portion of techniques feature; And these amendments or replacement, do not make the essence of appropriate technical solution depart from the spirit and scope of various embodiments of the present invention technical scheme.

Claims (6)

1. the space-time analysis method of urban road traffic state, it is characterized in that, the method comprises the steps:

1) from database, read the road network traffic flow space-time data needed for computational analysis, and be stored in ephemeral data table, set up space-time foundation data system;

2) data exception identification and processing rule is utilized to identify the exception that described traffic flow space-time data exists and repair;

3) from GIS map, obtain the Space Lorentz Curve matrix between road, binding time syntople, obtain space-time adjacency matrix;

The concrete grammar that described space-time object adjacent relation matrix calculates is as follows:

According to definition, be the space-time neighborhood of this space-time object with all space-time object on a space-time object adjacent space position, adjacent time point, so, space-time syntople just can be determined by Space Lorentz Curve and sequential syntople; Because adjacent weight is a value of getting 0 or 1, get 1 expression and adjoin, get 0 expression and do not adjoin, therefore space-time neighboring rights weight values can be expressed as the product of Spatial Adjacency weighted value and time neighboring rights weight values; If space-time object ST (p, i) and ST (q, j) is section p in the state in i moment and the section q state in the j moment respectively, the adjacent weight so between space-time object ST (p, i) and ST (q, j) is: w _{(p, i) (q, j)}=w _{(p, q)}× w _{(i, j)}

Wherein w _{(p, q)}spatial Adjacency weight between section p and section q:

W _{(i, j)}for the time between moment i and moment j adjoins weight:

Specify the space-time neighborhood each other of same moment of adjacent space position herein, but not space-time object each other between the adjacent moment of same locus, this is because the state of the adjacent moment of the same space geographic object is a time series, this time autocorrelation may cover special heterogeneity, therefore, same section p=q not spatial neighborhood is each other specified in computing formula, but synchronization i=j time neighborhood each other;

Calculate all w _{(p, i) (q, j)}, obtain a space-time object adjacency matrix, this step calculates the space-time object adjacency matrix obtained, the traffic flow space-time data after repairing with previous step, inputs next step together and carries out space-time auto-correlation index calculate;

4) according to the road network traffic flow space-time data carrying out repair process in step 2, and the space-time object adjacency matrix that step 3 obtains, and space-time auto-correlation index computation model, calculate the autocorrelative general indices of space-time and local indexes;

Calculating concrete grammar and the process of described space-time auto-correlation overall situation and partial situation index are as follows:

If y _{(p, i)}be a certain property value of the state in space-time object ST (p, i), p section i moment, so in certain hour section and spatial dimension, all space-time object can be expressed as about the overall auto-correlation index A of this property value:

$A=\frac{\mathrm{NT}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^N{w}_{(p,i)(q,j)}(y_{(p,i)}-\stackrel{\&OverBar;}{y})(y_{(q,j)}-\stackrel{\&OverBar;}{y})}{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}^2\&times;{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

Wherein N be research spatial dimension in the number in all sections, T be research time range in the time interval number of data acquisition, NT is the number of all space-time object, w _{(p, i) (q, j)}for the neighboring rights weight values between space-time object ST (p, i) and ST (q, j), average for all space-time object y property values:

$\stackrel{\&OverBar;}{y}=\frac{1}{\mathrm{NT}}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{y}_{(p,i)}$

The value of overall situation auto-correlation index A, in-1 to 1 scope, is in the implication that different sub-ranges represents different:

A is greater than 0 expression positive correlation, more more obvious close to 1 this trend, is embodied in studied property value and reaches unanimity in adjacent segments adjacent moment;

It is uncorrelated that A equals 0 expression;

A is less than 0 expression negative correlation, and more close-1 this trend is more obvious, is embodied in the value of studied property value in adjacent segments adjacent moment and is tending towards contrary;

Overall situation auto-correlation index A is the index reflecting overall space-time object auto-correlation situation, the autocorrelative local characteristics of space-time cannot be weighed, for concrete some space-time object, the correlationship in it and its space-time neighborhood between all space-time object is weighed, i.e. space-time auto-correlation local indexes by another index:

a _(p,i)＝Z _(p,i)Wz _(p,i)

Wherein: ${\mathrm{Wz}}_{(p,i)}=\frac{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}{Z}_{(q,j)}}{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

$Z_{(p,i)}=\frac{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}{\mathrm{\&sigma;}}$

$\mathrm{\&sigma;}=\sqrt{\frac{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}^2}{\mathrm{NT}-1}};$

5) multi-angle displaying and assistant analysis are carried out, to obtain urban road traffic state to the space-time auto-correlation index calculate result drawn in described step 4.

2. the space-time analysis method of urban road traffic state according to claim 1, is characterized in that, reads traffic flow space-time data and comprise in described step 1 from database:

101) querying condition is arranged to parameter arranged by user individual, described querying condition comprises from date and time, date of expiry and time, road network scope, described road network scope allows user directly to select from map, or directly inputs section title or section numbering in a database;

102) connection data storehouse, is nested in input database in query statement carries out the query manipulation of data by the querying condition that user inputs, and inquiry the data obtained leaves in an interim tables of data, as the basic data of follow-up anomalous identification and process operation.

3. the space-time analysis method of urban road traffic state according to claim 1, is characterized in that, the exception existed data in described step 2 identify and the concrete steps of repairing as follows:

201) data in the ephemeral data table that produces of read step 1, judge the missing data that exists in data and misdata according to the recognition rule of abnormal data, row labels of going forward side by side, as the foundation that follow-up abnormal data reparation operates;

202) according to the anomalous identification result that disorder data recognition process markup draws, the reparation rule according to abnormal data is repaired abnormal data, repairs complete data and is input to the calculating that subsequent step carries out auto-correlation index.

4. the space-time analysis method of urban road traffic state according to claim 1, is characterized in that, the concrete steps obtaining space-time adjacency matrix in described step 3 are as follows:

301) syntople between the spatial object obtaining two space-time object places, a certain bar section that what spatial object herein referred to is exactly spatially, two sections have common node then to represent they are adjacent, otherwise non-conterminous;

302) obtain two space-time object places time object between neighbouring relations, herein because traffic flow space-time data all gathers by certain time interval on section, therefore the time range of research is divided into a lot of periods by data collection interval, if two times in adjacent time interval, then show that two time objects are adjacent;

303) according to the space-time syntople between the Space Lorentz Curve obtained and time syntople determination space-time object, if two space-time object spatially with the time on all adjacent, so they are just adjacent in time-space relationship; Any two space-time object are all drawn to their syntople by above-mentioned steps, space-time adjacency matrix can be obtained.

5. the space-time analysis method of urban road traffic state according to claim 1, is characterized in that, in described step 4, space-time auto-correlation index computation model specifically comprises as follows:

401) calculating of overall auto-correlation index, overall situation auto-correlation index weighs measuring of all space-time object auto-correlation degree, what investigate is the auto-correlation relation existed in studied traffic state data between space-time object, can react the auto-correlation degree of all space-time object intuitively;

402) calculating of local auto-correlation index, local auto-correlation index weighs of auto-correlation degree between all space-time object in concrete some space-time and its space-time neighborhood to measure, investigation be auto-correlation relation in concrete some space-time object and its space-time neighborhood between all space-time object on studied traffic flow attribute space-time data.

6. the space-time analysis system of urban road traffic state, is characterized in that, this system comprises:

Traffic flow spatiotemporal data warehouse module, for realizing the real-time query function of traffic flow space-time data, and owing to needing data volume to be processed very large, and analytic process needs to carry out for different time sections, different local road network, requirement on flexibility is very high, and therefore described data inquiry module can according to real needs for follow-up statistical study provides basic data;

The identification of abnormal data and reparation module, for the traffic flow space-time data drawn traffic flow spatiotemporal data warehouse module, carry out identification and the reparation of abnormal data, because detecting device is malfunctioning, transmission line failure reason, inevitably there is mistake, deletion condition in the data in database, this module processes these abnormal datas, prevents the accuracy of abnormal data impact analysis result;

The computing module of space-time auto-correlation index, for receiving the traffic flow space-time data after carrying out disorder data recognition and repair process, according to the Space Lorentz Curve obtained by GIS map and time syntople determination space-time adjacency matrix, then by space-time adjacency matrix input space-time auto-correlation index computation model, the calculating of space-time auto-correlation index is carried out;

Calculating concrete grammar and the process of described space-time auto-correlation overall situation and partial situation index are as follows:

If y _{(p, i)}be a certain property value of the state in space-time object ST (p, i), p section i moment, so in certain hour section and spatial dimension, all space-time object can be expressed as about the overall auto-correlation index A of this property value:

$A=\frac{\mathrm{NT}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^N{w}_{(p,i)(q,j)}(y_{(p,i)}-\stackrel{\&OverBar;}{y})(y_{(q,j)}-\stackrel{\&OverBar;}{y})}{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}^2\&times;{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

Wherein N be research spatial dimension in the number in all sections, T be research time range in the time interval number of data acquisition, NT is the number of all space-time object, w _{(p, i) (q, j)}for the neighboring rights weight values between space-time object ST (p, i) and ST (q, j), average for all space-time object y property values:

$\stackrel{\&OverBar;}{y}=\frac{1}{\mathrm{NT}}{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{y}_{(p,i)}$

The value of overall situation auto-correlation index A, in-1 to 1 scope, is in the implication that different sub-ranges represents different:

A is greater than 0 expression positive correlation, more more obvious close to 1 this trend, is embodied in studied property value and reaches unanimity in adjacent segments adjacent moment;

It is uncorrelated that A equals 0 expression;

A is less than 0 expression negative correlation, and more close-1 this trend is more obvious, is embodied in the value of studied property value in adjacent segments adjacent moment and is tending towards contrary;

Overall situation auto-correlation index A is the index reflecting overall space-time object auto-correlation situation, the autocorrelative local characteristics of space-time cannot be weighed, for concrete some space-time object, the correlationship in it and its space-time neighborhood between all space-time object is weighed, i.e. space-time auto-correlation local indexes by another index:

a _(p,i)＝Z _(p,i)Wz _(p,i)

Wherein: ${\mathrm{Wz}}_{(p,i)}=\frac{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}{Z}_{(q,j)}}{{\mathrm{\&Sigma;}}_{q=0}^N{\mathrm{\&Sigma;}}_{j=0}^T{w}_{(p,i)(q,j)}}$

$Z_{(p,i)}=\frac{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}{\mathrm{\&sigma;}}$

$\mathrm{\&sigma;}=\sqrt{\frac{{\mathrm{\&Sigma;}}_{p=0}^N{\mathrm{\&Sigma;}}_{i=0}^T{(y_{(p,i)}-\stackrel{\&OverBar;}{y})}^2}{\mathrm{NT}-1}};$

Result of calculation assistant analysis and display module, for the space-time auto-correlation overall situation and partial situation index obtained the computing module of space-time auto-correlation index, carry out patterned Dynamic Display, draw trend broken line graph from time dimension and space dimension angle, distribution scatter diagram, histogram and traffic behavior show thematic maps, obtain traffic state space-time Evolution.