Classifications

• • G—PHYSICS
  • G08—SIGNALLING
  • G08G—TRAFFIC CONTROL SYSTEMS
  • G08G1/00—Traffic control systems for road vehicles
  • G08G1/01—Detecting movement of traffic to be counted or controlled
  • G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions

Definitions

• Method for forecasting a parameter representing the state of a system in particular a traffic parameter representing the state of a traffic network, and device for carrying out the method
• the invention relates to a method for forecasting a parameter representing the state of a system, in particular a traffic parameter representing the state of a traffic network, and a device for carrying out the method.
• a forecast of a traffic parameter relating to the state of a traffic network for a future point in time can take place taking into account the time-periodic courses of this parameter.
• the periodic courses of the traffic parameter also called aisle lines, can be obtained from traffic data for this traffic parameter at different times by statistical compression.
• a curve (ie a course) of a traffic parameter can be, for example, the course during the time of day of a certain day of the week, during a week and / or during the year.
• Traces of traffic parameters that are compressed and stored as curve lines can be provided with selection features so that a forecast is possible by comparing, for example, the current situation with at least one selection feature of at least one curve line.
• One problem with this is that the current situation with regard to a selection parameter is one
• the gait line does not indicate the future course of the traffic parameter represented by this gait line that is to be predicted with sufficient reliability.
• the object of the present invention is the most efficient possible optimization of forecasts, in particular traffic forecasts.
• the object is solved by the subject matter of the independent claims.
• a method according to the invention optimizes forecasts of parameters, in particular traffic parameters.
• Traffic parameters of a traffic network can be forecast in high quality on the basis of data relating to a second parameter of the system and at least one aisle line. This is particularly advantageous in cases in which the future course of a traffic parameter to be predicted can be better deduced on the basis of current values of another traffic parameter than on the basis of the current values of the former traffic parameter. If, for example, a forecast for the car travel times in the late morning is to be made in the early morning, the current car travel times are unsuitable for a forecast of the future car travel times, since there are hardly any cars driving early in the morning on weekdays or at weekends, but this is not a statement about cars driving late in the morning.
• Such a method could also be referred to as coupled melting (statistical compression of actual courses of parameters) and probabilistic selection.
• coupled melting statistic compression of actual courses of parameters
• probabilistic selection In order to obtain movement curves (courses) of traffic parameters, actual courses are examined and together with
• Selection characteristics are saved. Furthermore, it is examined which dependencies exist between different traffic parameters in order to enable a prediction of a first parameter on the basis of data on a second parameter.
• the fixed or time-dependent strength of couplings in each case at least two parameters is preferably examined and stored with. It is also possible to update the key figure representing the coupling strength of at least two parameters on the basis of current actual courses of the parameters and / or the quality of forecasts. Couplings of various parameters are also taken into account in the stored data on the curve. It is also advantageous to take into account and store the variance (or variability) of the courses of a parameter condensed into a curve and to take the variance (or variability) into account when forecasting a parameter.
• a probabilistic selection of a curve can consist in taking into account the probability that a certain curve is due to a measurement of the second parameter in order to forecast a parameter for a future point in time based on data for another parameter at the current point in time for the selection of a curve for forecasting good forecast for the first
• a self-correction of the curve base is preferably carried out by carrying an error curve, in which deviations from predicted courses from actual courses are taken into account for the correction of course curves.
• a continuous correction of the key figures for the coupling strength of at least two parameters is also expedient; Large deviations from actual values to predicted values in particular can lead to a weakening, small deviations of the actual values from the predicted values can lead to a coupling being strengthened.
• the method can in particular be implemented as a program in a traffic control center; in the traffic control center can in particular comprise a database with corridors (courses of traffic-related or other parameters) and / or a database with key figures for coupling at least two parameters each.
• Fig. 1 as a block diagram the statistical compression (melting) of
• Fig. 2 shows an example of a forecast of a parameter based on current
• traffic data 1 from floating cars (FCD), traffic data 2 from above-ground detectors (SES data) and traffic data 3 from induction loop data (VIZ) are measured at several locations at several points in time at one location, for example, one of the two courses 5, 6 of a traffic parameter shown as an example in box 4 can result.
• step 5 the courses (5, 6, etc.) of FCD data 1, SES data 2, VIZ data 3 are melted coupled, that is, taking into account couplings
• Characteristic curves, characteristic curve-related selection characteristics and couplings between key figures representing traffic parameters are summarized statistically and stored in a forecast database. For example, the course of the number of cars in a section of a route on a working day, the course of the number of cars in a section on a weekend, the course of the number of trucks in a section on a working day, the course of trucks on a section on a Sunday are each statistically compressed into a separate curve (chronological progression on a weekday at a position) and provided with selection features. Selection features can be, for example, the number of cars at a certain time and the number of trucks at one certain time etc. Selection characteristics are assigned to at least one or possibly also several curve lines.
• a selection characteristic or several selection characteristics of a gear line are currently fulfilled, for example if the number of trucks is currently (early in the morning) above a certain value, it can be concluded that a specific gear line (truck / working day) is currently being tracked. From this, a forecast can be made for the traffic parameter belonging to the measured data at a future time or according to the invention for a traffic parameter not assigned to the measured data at a future time. The curve of a traffic parameter that is most likely to be the future based on measured current data
• Figure 2 shows an example of a probabilistic selection.
• traffic data on the current number of cars and the current number of trucks in a section of road are available.
• the number of cars in a section of road for a future point in time, namely late in the morning, is to be forecast. Due to the current (early morning) number of cars, this is not possible, since the gangways of cars hardly differ on weekdays and early in the weekend.
• the number of trucks in a course for a course of a working day and a course for a course of Sunday clearly differ early in the morning. Based on the number of trucks in the morning on a workday aisle, it can therefore be concluded that the number of cars on a workday aisle will continue to develop and that therefore in the late morning the number of cars on the car aisle is applicable for late mornings.
• the number of cars on a workday aisle will continue to develop and that therefore in the late morning the number of cars on the car aisle is applicable for late mornings.
• the coupling can be taken into account in binary or quantized form. If several
• the most likely pathway can be selected.
• the method was developed to forecast traffic parameters.
• another parameter can also be predicted according to the invention. For example conclude from the morning flow of cars the concentration of pollutants at noon etc.

Landscapes

• Chemical & Material Sciences (AREA)
• Analytical Chemistry (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Traffic Control Systems (AREA)
• Data Exchanges In Wide-Area Networks (AREA)
• Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Applications Claiming Priority (3)

Publications (2)

Family

ID=7850246

Family Applications (1)

Country Status (4)

Families Citing this family (12)

Family Cites Families (3)

• 1997
  • 1997-11-18 DE DE19753034A patent/DE19753034A1/de not_active Withdrawn
• 1998
  • 1998-09-25 EP EP98958804A patent/EP1032927B1/fr not_active Expired - Lifetime
  • 1998-09-25 WO PCT/DE1998/002932 patent/WO1999026210A1/fr active IP Right Grant
  • 1998-09-25 AT AT98958804T patent/ATE235729T1/de active
  • 1998-09-25 DE DE59807678T patent/DE59807678D1/de not_active Expired - Lifetime

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events