EP0992031A1

EP0992031A1 - Method for determining travel time according to the time of the day

Info

Publication number: EP0992031A1
Application number: EP98939482A
Authority: EP
Inventors: Karla Oberstein; Thomas Schuster
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 1997-06-24
Filing date: 1998-06-24
Publication date: 2000-04-12
Also published as: WO1998059329A1

Abstract

The invention relates to a method for determining travel time according to the time of the day in travel guidance and information systems requiring travel times for optimal navigation purposes for given route sections according to the time of day. When determining the travel times of vehicles (1) traveling in a route network, only a few travel times are known in the introductory phase. Enhancement by means of digital maps is often too imprecise. According to the inventive method, the daily schedule is divided into time periods to which measured travel times are allocated. Average travel time values (RZ) are determined by communicating travel times which are provided for by various vehicles. Once the daily schedule is completed, the measured data is interpreted by determining a basic travel time (B), by eliminating highly divergent average travel time values (RZ) and by determining substitute travel times for time interval (ZI) without average travel times (RZ) on the basis of basic travel time (B). The curve thus obtained is subsequently smoothed out.

Description

description

Procedure for determining a travel time dependent on the time of day

The invention relates to a method for determining a travel time dependent on the time of day according to the preamble of patent claim 1.

Drivers should be offered dynamic route recommendations to their desired destination. To do this, it is necessary to know the current travel times on sections of the route in a busy road network.

On the one hand, systems with loop detectors are used to determine travel times, which produce a large number of measured values at a few, specifically selected points (for example traffic signal systems).

Another system is known from EP 0292897, which calculates the travel times of individual vehicles for specific ones

Route sections allowed by direct communication with a traffic control center. Particularly when introducing such systems, however, only a few vehicles are initially equipped for it, so that, in contrast to the systems with loop detectors, there are few measured values on many different sections of the route. As a result, the current travel times are not known at all times.

A method is known from a dissertation (Dissertation, Dr. Janko, "Problems of Travel Time Prediction in a Control System for Road Traffic", Department 10: Transportation and Applied Mechanics, TU Berlin, 1994), in which the travel times transmitted by the vehicles each take five minutes , time intervals covering an entire daily routine can be assigned. The transmitted travel times are initially offset against travel times of neighboring time intervals and with travel times known from digital maps before they are assigned to the individual time intervals. Travel times derived from street categories of digitized maps generally do not correspond to the travel times actually traveled on these routes. The immediate settlement with travel times of temporally adjacent time intervals also leads to erroneous travel time forecasts due to the resulting strong influence of individual incorrect measurements in neighboring time intervals.

The object of the present invention is to compare one in a traffic management and information system with measurement of the travel time by individual vehicles, with a few measured travel times present, as is the case, for example, in the introduction phase of such systems or on low-traffic routes State of the art to create improved travel time forecast.

The object is achieved by a method having the characterizing features of claim 1.

The method according to the invention advantageously saves travel times actually traveled. At the beginning of each day, the measured travel times of the previous days are offset against a constant dependence of the travel time on the time of day. Only the constant dependency allows the determination and transmission of optimal route data to vehicles for any route at any time of the day. The travel times actually traveled remain as an unadulterated data record and, together with the travel times measured for the current day, form the basis for the calculations for the next day. As a result, the determined time of day travel time is becoming more and more accurate, and outliers can also be identified at later times. In contrast to this, in the known method, through the immediate manipulation of the measured travel times by comparison with route categories and with travel times of temporally adjacent intervals, the actually measured travel times are falsified, so that further measured values always have to be compared with already manipulated travel times. Therefore, the known method does not ensure that the travel times become more accurate over time.

Advantageous refinements of the method for determining the replacement travel times are described in the dependent claims. Replacement travel times are determined on the one hand from a so-called basic travel time, which essentially results from averaging the measured travel times over the entire daily routine. For time intervals without assigned actually measured travel times, in the time environment of which there are time intervals with assigned measured travel times, the measured travel times of the neighboring time intervals are also taken into account in the calculation of the substitute travel times. In contrast to the known method, the replacement travel times are determined from actually measured travel times without taking into account the street categories from digitized maps.

In the following, the method is explained in more detail with the help of illustrations, showing

1 shows the basic structure of a system for measuring travel time in traffic control systems, FIG. 2 shows a graphical representation of the assignment of average travel time values to time intervals,

FIG. 3 shows a graphic representation of the formation of alternative travel times for time intervals to which no average travel time values can be assigned on the basis of measurements, and FIG. 4 shows a graphic representation of a smooth, constant dependence of the travel time on the time of day (curve), which is obtained from the data in FIG. 2 has been .

FIG. 1 shows schematically how a vehicle 1 uses a travel time measuring device 2, for example a wheel pulse generator, to measure the travel time on a section of the route, which is shown on a digitized map that is displayed in a navigation system. gationsrechner 3 is stored in the vehicle 1, as a route between two traffic nodes of a meshed road network is defined. The measured travel time is transmitted from the navigation computer 3 via a transmitting and receiving device 4 to the transmitting and receiving device 5 of a traffic computer 6 of a traffic control center 7.

In a traffic management and information system, motorists are to receive a route recommendation from a traffic control center 7 in order to reach their destination, which was determined as far as possible using current travel times on the route sections belonging to the route. Nominally longer secondary routes, which are used less at the time of the request and thus can be traveled more quickly, should be taken into account when determining the route recommendation by the traffic control center 7. For such a route recommendation, it is therefore necessary to have the required travel times of all route sections available in the traffic control center at all times of the day. To ensure that route recommendations can be transmitted to vehicles even in the introductory phase of a traffic control and information system, when only a few vehicles are equipped with the appropriate devices and route section-specific travel times are not yet available at all times of the day, a procedure that is explained in more detail below is used uses the few available measured travel times to determine a travel time dependent on the time of day for all route sections.

For this purpose, a daily routine is divided into time intervals ZI, for example 288 consecutive five-minute time intervals ZI, and in the traffic computer 6 the travel times transmitted are initially assigned to the respective time intervals ZI and route sections in which they were transmitted. Outliers with extremely low or high travel times that clearly indicate an incorrect measurement or a parking of the vehicle are eliminated - regardless of the route category. For each time interval ZI, an average travel time RZ is determined in the first step of the method by averaging travel times transmitted by different vehicles, which leads to an assignment between average travel time values RZ and time intervals ZI, as shown in FIG. 2 for a route section. Strong fluctuations will generally occur between the travel times of adjacent time intervals and an assigned travel time mean value RZ has not been determined for each time interval ZI. In order that a route recommendation to a vehicle can be determined, a substitute travel time is determined for each time interval ZI without an assigned travel time mean value RZ and the fluctuations are then reduced by smoothing.

For this purpose, in the second step of the method, a so-called base travel time B is determined for each section of the route, which corresponds to a travel time at times with little traffic. For sections of the route on which there are no entries at all or fewer than a specified number, the basic travel time B must be specified from the route categories of digitized maps. The base travel times B determined from this often do not match the actual travel times on the route sections. From just a few measured

Travel times and the travel time mean values RZ determined therefrom, an improved base travel time B is to be determined, for example by calculating the median of all travel time mean values RZ of the route section.

In the third step of the method, singular travel time mean values RZ are eliminated, that is to say travel time mean values RZ which have an unusually long or particularly short travel time, which is determined, for example, by determining the statistical standard deviation of all travel time mean values RZ des _. Route section and subsequent division of the travel time mean values RZ into three areas is possible: A first The travel time average values RZ, which have a time interval from the basic travel time B between three and four times the standard deviation, are assigned to the area, a second area are assigned to travel time average values RZ, which have a time interval from the basic travel time B between four times and five times the standard deviation, and a third area the travel time mean values RZ are assigned, which have a time interval from the basic travel time B greater than five times the standard deviation. If one of these areas is not occupied, then all travel time mean values RZ which lie in a subsequent area are eliminated, which means that time intervals ZI with such singular travel time mean values RZ no longer have any assigned travel time mean values RZ.

In the fourth step, alternative travel times are determined for time intervals ZI which do not have assigned travel time mean values RZ after the third step or because of a lack of transmitted travel times for this time interval ZI.

It should be noted that, especially in times of low traffic, for which the base travel time B is a good approximation of the travel time, few travel times are measured, since there are few vehicles 1 with travel time measuring devices 2. If, for example, a travel time is measured on a section of the route only during rush hour traffic at 7:00 a.m. and 4:00 p.m., which is far above the basic travel time B due to rush hour traffic, experience has shown that ZI does not apply to all time intervals between 7:00 a.m. and 4:00 p.m. a similarly long travel time is also to be expected, so that linear interpolation is not a suitable method for determining alternative travel times. With a high measured travel time at 7:00 a.m. and a further one at 7:20 a.m. which allows a linear interpolation between the measured travel times.

By choosing both a small number n1 of time intervals ZI, which defines a small time period, and a large number n2 of time intervals ZI, which defines a long time period, a time period without measured travel times is divided into three classes, that are treated differently. If the time without measured travel times is shorter than the short time, experience shows that the traffic on the route section changes little and linear interpolation is carried out. If the time period without measured travel times is greater than the large time period, experience has shown that the travel time after the large time period has decreased linearly to the basic travel time B, for the middle range of a time period without measured travel times of the small and large time periods Base travel time B is taken into account when determining the alternative travel times.

For this, depending on the number of successive time intervals ZI ZI without assigned travel time mean values RZ for the small number nl and the large number n2 of time intervals ZI, one of the following method steps is carried out:

If the number NZI of the successive time intervals ZI without travel time average values RZ is smaller than the small number nl of time intervals ZI, then the alternative travel times to be assigned to the successive time intervals ZI without travel time average values RZ are formed by linear interpolation. This is carried out between a last travel time average R1, assigned to the last time interval IA, which is before the successive time intervals without travel time average values RZ, and a next travel time average R2, assigned to the next time interval IN, which is temporally after the successive time intervals ZI without travel time mean RZ.

If the number NZI of the successive time intervals ZI without assigned travel time mean values RZ is greater than the large number n2 of time intervals ZI, then the basic travel time B is assigned to a time interval ZI which has a time interval corresponding to the large number n2 of time intervals from the last time interval IA , and then the substitute travel times, which are to be assigned to the time intervals ZI between the time interval with the assigned basic travel time B and the last time interval IA, are formed by linear interpolation between the basic travel time B and the last travel time mean value R1.

If the number NZI of the successive time intervals ZI without assigned travel time mean values RZ is greater than the small number nl but smaller than the large number n2 of time intervals ZI, then a time interval IM in the middle of the successive time intervals ZI without travel time mean values RZ becomes a special replacement travel time R5 assigned by the following calculation rule, which is shown in FIG. 3:

R5 =% x (R3 + R4), with R3 = Rl - (NZI) x (Rl - B) / (2 x n2), and R4 = R2 - (NZI) x (R2 - B) / (2 x n2 ).

Then the replacement travel times, the time intervals ZI between the last time interval IA and the

Time interval IM in the middle are to be determined by linear interpolation between the last travel time mean R1 and the special replacement travel time R5, and the replacement travel times which are to be assigned to the time intervals ZI between the time interval in the middle IM and the next time interval IN , by linear interpolation between the special actual replacement travel time R5 and the next travel time average R2 determined.

In the fifth step, the curve shape is smoothed in such a way that an approximately continuous dependence of the travel times on the time intervals ZI is recorded in the traffic computer 6 and as is shown schematically in FIG.

If another vehicle now asks the traffic computer 6 for an optimal route to its destination, the shortest distances and possible detours are checked for the shortest travel time based on the time of day travel times determined for the route based on the method, and the time-optimized route is transmitted to the further vehicle.

Claims

claims

1.Procedure for determining a time of day dependent on travel time on route sections of a road network, i.e. during defined time periods (ZI) that split up the duration of a day without gaps, characterized in that measured travel times of individual vehicles (1) belonging to the route sections are recorded and the Time intervals (ZI) in which they were measured are assigned that travel time mean values (RZ) per time interval (ZI) are determined by averaging the travel times assigned to a time interval (ZI) of different vehicles (1) so that a replacement travel time can be assigned to an individual Time interval depending on average travel time values (RZ) is assigned to neighboring time intervals (ZI), if this individual time interval is not assigned an average travel time value (RZ), and that the average travel time values (RZ) and the replacement travel times are averaged over several time intervals (ZI), to store approximately constantly changing travel times per time interval (ZI) in the traffic computer (6).

2. A method for determining a travel time dependent on the time of day according to claim 1, characterized in that with a travel time measuring device (2) connected to a navigation computer (3) in the vehicle (1), the travel time is measured on the route sections and with a transmission connected to the navigation computer (3) - and receiving device (4) in the vehicle (1) the measured travel time is transmitted to a transmitting and receiving device (5) of a traffic computer (6) of a traffic control center (7).

3. Method for determining a travel time dependent on the time of day according to one of claims 1 or 2, characterized in that that travel times that lie outside of previously defined limits are eliminated.

4. Method for determining a travel time dependent on the time of day according to one of claims 1, 2 or 3, d a d u r c h g e k e n n z e i c h n e t that a time of day independent base time (B) for the route section is formed from the average travel time values (RZ).

5. A method for determining a travel time dependent on the time of day according to claim 4, so that the base travel time (B) is determined by determining the median of all travel time mean values (RZ).

6. A method for determining a travel time dependent on the time of day according to claim 4 or 5, so that individual travel time mean values (RZ) with extreme deviations from the basic travel time (B) are eliminated.

7. Method for determining a travel time dependent on the time of day according to one of claims 4, 5 or 6, characterized in that the standard deviation of the travel time average values (RZ) is determined and the average travel time values (RZ) are determined in order to eliminate individual travel time mean values (RZ) with extreme deviations from the base travel time divided into three areas, the first area containing the travel time averages (RZ), which have a time interval between the three and four times the standard deviation from the basic travel time (B), the second area contains the travel time averages (RZ), which is a time interval between of four and five times the standard deviation from the basic travel time (B) and the third area contains the travel time mean values (RZ), which are a time interval greater than five times Standard deviation from the basic travel time (B) and, if one of these three areas is not occupied, all travel time averages (RZ) located in the following areas are eliminated.

8. A method for determining a travel time dependent on the time of day according to one of claims 4 to 7, characterized in that the substitute travel times are formed as follows: as a function of a comparison of the number (NZI) of successive time intervals, to which no average travel time value (RZ) is assigned, with a Predetermined small number (nl) and with a large number (n2) of time intervals (ZI) different methods are used:

if the number (NZI) of the successive time intervals is smaller than the small number (nl) of time intervals (ZI), then the substitute travel times to be assigned to the successive time intervals are formed by linear interpolation between a last travel time mean value (Rl) and assigned to the last time interval ( IA), which is before the successive time intervals, and a next average travel time (R2), assigned to the next time interval (IN), which is after the successive time intervals.

If the number (NZI) of the successive time intervals without an assigned travel time mean value (RZ) is greater than the large number (n2) of time intervals (ZI), then a time interval which is a time interval corresponding to the large number (n2) of time intervals becomes the last Time interval, the base travel time (B) assigned, and then the replacement travel times, which are to be assigned to the time intervals between the time interval with the assigned base travel time (B) and the last time interval (IA), by linear interpolation between the base travel time (B) and the last travel time mean (Rl) formed, if the number (NZI) of the successive time intervals without assigned travel time mean values (RZ) is greater than the small number (nl) but smaller than the large number (n2) of time intervals, then a time interval (IM) in the middle of the successive time intervals becomes a special one Substitute travel time (R5) is assigned, and then the substitute travel times to be assigned to the time intervals (ZI) between the last time interval (IA) and the time interval (IM) in the middle are determined by linear interpolation between the last travel time mean value (Rl) and the special one Substitute travel time (R5) is determined, and the substitute travel times that are to be assigned to the time intervals (ZI) between the time interval in the middle (IM) and the next time interval (IN), by linear interpolation between the special substitute travel time (R5) and the next travel time mean value ( R2) determined.

9. A method for determining a travel time dependent on the time of day according to claim 8, characterized in that the special replacement travel time (R5) by the following calculation rule from the number (NZI) of the successive time intervals, from the last travel time average (Rl), the next travel time average ( R2), the base travel time (B) and the large number (n2) of time intervals is determined: R5 = ^ x (R3 + R4), with R3 = Rl - (NZI) x (Rl - B) / (2 x n2) , and R4 = R2 - (NZI) x (R2 - B) / (2 x n2).