CN107331149B

CN107331149B - Method and device for predicting vehicle running time

Info

Publication number: CN107331149B
Application number: CN201610281163.0A
Authority: CN
Inventors: 杨和东; 耿璐
Original assignee: Hitachi Ltd
Current assignee: Hitachi Ltd
Priority date: 2016-04-29
Filing date: 2016-04-29
Publication date: 2021-07-09
Anticipated expiration: 2036-04-29
Also published as: CN107331149A

Abstract

The invention provides a method and a system for predicting vehicle running time, wherein the method comprises the following steps: a prediction model creation step (S1) of collecting historical travel information records from a start point to a target point from a historical travel information record library on the basis of a designated start point and target point, calculating a reference historical travel time of the historical travel information records using the historical travel information records, and determining at least one main influence factor having the greatest influence on the travel time among the travel influence factors to create a prediction model; and a vehicle travel time prediction step (S2) of, when it is necessary to predict the vehicle travel time from the start point to the target point, predicting the vehicle travel time from the start point to the target point using the prediction model that has been established, based on current actual input conditions including the start point as a vehicle departure place, the target point as a vehicle destination, and the main influence factor.

Description

Method and device for predicting vehicle running time

Technical Field

The invention relates to the field of vehicle scheduling, in particular to a method for predicting vehicle running time offline by using a decision tree algorithm and a device for predicting vehicle running time.

Background

In the field of vehicle scheduling, such as bus scheduling and logistics scheduling, it is very important to accurately predict the travel time between two points of interest. The travel time prediction uses a travel trace record described by GPS data and a time stamp collected by a terminal of a vehicle to accurately predict the travel time, improving scheduling efficiency and quality. When a specific route is given, the travel time tends to have only a lower boundary and no upper boundary, because the travel speed of the vehicle is affected by various predictable external factors such as weather conditions forecasted by a third party, traffic congestion conditions mined from traffic data, and the like, and also by low-frequency unpredictable external factors such as vehicle failure, the driver's own cause, and the like.

Vehicle travel time prediction can be classified into real-time prediction and offline prediction. Patent document 1 and patent document 2 disclose real-time prediction methods based on distance and current speed. Patent document 3 discloses an offline prediction method that utilizes a fuzzy function of a specific road segment within a city during a time window. Patent document 4 discloses a method of predicting in real time using an average value of historical travel data.

Documents of the prior art

Patent document

Patent document 1: chinese patent No. CN201410453120

Patent document 2: chinese patent No. CN201410211385

Patent document 3: chinese patent No. CN201410077956

Patent document 4: chinese patent No. CN201210259408

Disclosure of Invention

Technical problem to be solved by the invention

However, the methods for predicting the vehicle travel time in real time as described in patent documents 1, 2 and 4 rely on real-time traffic information, and when the travel times of many vehicles need to be predicted at the same time, the amount of data processed in real time is very large, which easily causes an excessive load on a processor, and a long wait is required to obtain a prediction result beyond the processing capability of the processor, thereby deteriorating the user experience.

On the other hand, the offline prediction method using the fuzzy function and the time window as described in patent document 3 and the method using the average value of the historical travel data for real-time prediction as described in patent document 4 take into account the influence of all the external factors, but when the influence of unpredictable external factors having low occurrence frequencies is taken into account, the occurrence frequencies of these unpredictable external factors are low, and the result of the travel time is an abnormal value, so that the method using the fuzzy function or the average value of the historical travel data as described above may reduce the accuracy of predicting the vehicle travel time in the normal case. Further, the methods described in patent documents 3 and 4 can be used only for short-distance prediction such as prediction of vehicle travel time in cities, and cannot be used for long-distance prediction such as long-distance distribution.

The present invention has been made in view of the problems of the prior art as described above, and an object thereof is to provide a method of predicting a vehicle travel time offline and an apparatus for predicting a vehicle travel time, which exclude the influence of an abnormal value.

Means for solving the technical problem

In order to achieve the above object, according to an aspect of the present invention, a method of predicting a travel time of a vehicle, includes: a prediction model establishing step of collecting historical travel information records from a starting point to a target point from a historical travel information record base according to a specified starting point and target point, calculating reference historical travel time of the historical travel information records by using the historical travel information records, and determining at least one main influence factor having the greatest influence on the travel time among the travel influence factors to establish a prediction model; and a vehicle travel time prediction step of, when it is necessary to predict a vehicle travel time from the start point to the target point, predicting the vehicle travel time from the start point to the target point using the prediction model that has been established, based on current actual input conditions including the start point as a vehicle departure place, the target point as a vehicle destination, and the main influence factor.

According to the method of predicting a vehicle travel time of the present invention as described above, since the influence of the abnormal value is excluded by using the reference historical travel time, the vehicle travel time can be predicted with high accuracy. In addition, the prediction model established offline according to the historical driving information record is adopted, when the user inputs the current actual driving condition, the driving time of the vehicle is predicted by using the prediction model, and the user does not need to rely on frequent and complicated input information such as real-time traffic data, so that the data volume to be processed by the processor in real time is greatly reduced, the load of the processor is reduced, the prediction result can be rapidly obtained, the prediction efficiency is improved, and the driving time prediction service can be provided for more vehicles at the same time. The method and the system can be used for short-distance prediction such as prediction of vehicle running time in cities and long-distance prediction such as long-distance logistics distribution.

The prediction model building step may include: a collection step of collecting a history travel information record from a start point to a target point from the history travel information record base according to a specified start point and target point, the history travel information record including history travel time and travel influence factors from the start point to the target point; a first reference historical travel time calculation step of calculating a reference historical travel time of the historical travel information record using the historical travel time, an arrival state determination step of determining an arrival state of the historical travel information record indicating whether the historical travel time is late compared to the reference historical travel time by comparing the historical travel time with the reference historical travel time; a main influence factor determination step of analyzing the historical travel information record based on the arrival state, thereby determining at least one main influence factor having the greatest influence on the arrival state among the travel influence factors; grouping, namely grouping the historical driving information records according to the main influence factors; and a second reference historical travel time calculation step of calculating the reference historical travel time of each group of the grouped historical travel information records.

Specifically, in the collecting step, n pieces of history travel information records from the start point to the target point, which include history travel time from the start point to the target point and k kinds of travel influence factors, where n and k are natural numbers, may be collected according to the designated start point and target point; in the first reference historical travel time calculating step, the reference historical travel time of the n historical travel information records as a whole may be calculated from the historical travel time in each of the n historical travel information records; in the arrival state determining step, the historical travel time of each of the n historical travel information records may be compared with a reference historical travel time of the n historical travel information records as a whole, respectively, to determine the arrival state of each of the n historical travel information records, the arrival state indicating whether the historical travel time of each of the n historical travel information records is late compared with the reference historical travel time of the n historical travel information records as a whole, and the state values of the arrival state include m kinds of arrival state values, where m is a natural number and 0 < m ≦ n; in the main influence factor determining step, the n historical driving information records may be analyzed as a whole by using a decision tree algorithm based on the determined arrival state of each of the n historical driving information records, so as to determine a main influence factor having the largest influence on the state value of the arrival state among the k driving influence factors, and divide the state value of the main influence factor into q state categories corresponding to the m arrival state values, where q is a natural number and 0 < q ≦ m; in the grouping step, the n historical travel information records may be divided into q groups of corresponding historical travel information records according to the state categories of the main influence factors of the n historical travel information records; and in the second reference historical travel time calculating step, the reference historical travel time for the entire set of historical travel information records may be calculated from the historical travel time for each historical travel information record in each of the q sets of historical travel information records.

Further, in the vehicle travel time prediction step, a set of historical travel information records corresponding to the state category of the main influence factor in the current actual input condition may be determined based on the current actual input condition, and a reference historical travel time of the set of historical travel information records may be predicted as the vehicle travel time.

And predicting the reference historical travel time of a group of historical travel information records corresponding to the state type of the main influence factor in the current actual input condition as the vehicle travel time. Therefore, the accuracy of predicting the vehicle travel time is improved, and when the user inputs the current actual travel condition, the prediction result can be determined by quickly finding the reference historical travel time of the corresponding set of historical travel information records without complicated calculation, thereby improving the prediction efficiency.

In the first reference historical travel time calculation step, a median of all the historical travel times of the n historical travel information records may be set as the reference historical travel time of the entire n historical travel information records. In the second reference historical travel time calculation step, a median of all the historical travel times in each set of the historical travel information records may be set as the reference historical travel time for the entire set of the historical travel information records.

Since the median of the historical travel time is used as the reference historical travel time, the influence of the abnormal value on the prediction result when prediction is performed, such as the average value, is eliminated, and the accuracy of predicting the travel time is further improved.

The method of predicting a vehicle travel time may further include: and a self-learning step of writing the actual travel time of the vehicle from the starting point to the target point into the historical travel information record in association with the starting point, the target point and the main influence factor each time the vehicle reaches the destination, so as to update the prediction model.

Therefore, the self-learning step updates the historical travel information record with the newly generated travel information record at regular intervals to update the prediction model, thereby further improving the accuracy of the predicted travel time.

According to another aspect of the present invention, an apparatus for predicting a travel time of a vehicle, includes: a prediction model establishing unit which collects historical driving information records from a starting point to a target point from a historical driving information record base according to a specified starting point and the target point, calculates reference historical driving time of the historical driving information records by using the historical driving information records, and determines at least one main influence factor which has the largest influence on the driving time in the driving influence factors to establish a prediction model; and a vehicle travel time prediction unit that predicts a vehicle travel time from the start point to the target point using the prediction model that has been established, when it is necessary to predict a vehicle travel time from the start point to the target point, based on current actual input conditions including the start point as a vehicle departure place, the target point as a vehicle destination, and the main influence factor.

According to the apparatus for predicting a vehicle travel time of the present invention as described above, since the influence of the abnormal value is excluded by using the reference historical travel time, the vehicle travel time can be predicted with high accuracy. In addition, the prediction model established offline according to the historical driving information record is adopted, and when the user inputs the current actual driving condition, the driving reality of the vehicle is predicted by using the prediction model, so that the data volume to be processed by the processor in real time is greatly reduced, the load of the processor is reduced, the prediction result can be obtained quickly, and the prediction efficiency is improved. The method and the system can be used for short-distance prediction such as prediction of vehicle running time in cities and long-distance prediction such as long-distance logistics distribution.

Effects of the invention

According to the present invention, it is possible to provide a method of predicting a vehicle travel time offline and an apparatus for predicting a vehicle travel time, which exclude the influence of an abnormal value.

Drawings

Fig. 1 is a block diagram showing a device for predicting a vehicle travel time according to a first embodiment of the present invention.

Fig. 2 is a functional block diagram showing a prediction model creating unit of the apparatus for predicting the vehicle travel time according to the first embodiment of the present invention.

Fig. 3 is a flowchart showing a method of predicting a vehicle travel time according to a first embodiment of the present invention.

Fig. 4 is a detailed flowchart showing a vehicle travel time prediction step in the method of predicting a vehicle travel time according to the first embodiment of the present invention.

Fig. 5 is a block diagram showing a device for predicting a vehicle travel time according to a second embodiment of the present invention.

Fig. 6 is a flowchart showing a method of predicting a vehicle travel time according to a second embodiment of the present invention.

Fig. 7 is an example table showing a vehicle history travel information record of the embodiment of the invention.

Fig. 8 is an exemplary table showing a vehicle history traveling information record after determining the arrival state of the vehicle according to the embodiment of the present invention.

FIG. 9 is an example of a process for determining primary influencing factors using a decision tree algorithm that represents an embodiment of the present invention.

Fig. 10 is an exemplary table showing the grouping of the history traveling information records after determining the main influence factor according to the embodiment of the present invention.

Detailed Description

Hereinafter, a detailed description will be given of a specific embodiment of the present invention with reference to the drawings. In the drawings, like reference characters designate the same or similar parts throughout the several views.

[ first embodiment ] to provide a liquid crystal display device

An apparatus and method for predicting a vehicle travel time according to a first embodiment of the present invention will be described with reference to fig. 1 to 4.

Fig. 1 is a block diagram showing a device 1 for predicting a vehicle travel time according to a first embodiment of the present invention. As shown in fig. 1, a device 1 for predicting vehicle travel time according to a first embodiment of the present invention includes a prediction model creation unit 20 and a vehicle travel time prediction unit 30.

The prediction model creation unit 20 is a unit for creating a prediction model to be utilized by the apparatus 1 that predicts the vehicle travel time. The prediction model creation unit 20 creates a prediction model as follows: from a history travel information recording library not shown, history travel information records from a start point to a target point are collected from designated start points and target points, a reference history travel time of the history travel information records is calculated using the history travel information records, and a main influence factor having the largest influence on the travel time among the travel influence factors is determined.

The historical travel information record base is a database in which historical travel information records of each vehicle traveling between an arbitrarily designated starting point and a target point in the past (for example, from city a to city B, from west gate to tribit bridge in beijing, and the like) are recorded, and when the prediction model creating unit 20 collects the historical travel information records of the starting point to the target point, it is preferable to collect the historical travel information records of the same route from the starting point to the target point. The history travel information record is data in which travel information of the vehicle is stored in association with vehicle information.

The history travel information record stores a start point, a target point, and a travel time from the start point to a target point of the vehicle in association with various travel-affecting factors that may affect the travel time. For example, the history travel information record stores departure date, departure time, vehicle type, load capacity, weather condition at the time of travel, whether or not a holiday is public, and actual travel time of each type of vehicle from the a site to the B site in association with each other.

The driving influence factor may include departure date, departure time, model type, load capacity, weather condition while driving, public holiday or not, etc. of the vehicle, but this is only an example, and only a part of the aforementioned items may be included, more items may be included, or the aforementioned items may be replaced with other items. The influence factors such as departure date and time of the vehicle may be generated from the time stamp collected by the vehicle terminal or may be input by the user, but are not limited thereto. Influence factors such as whether the public holidays are available or not, weather conditions and the like can be received through the vehicle network and sent to the historical driving information recording base to be recorded in the historical driving information record, but the method is not limited to the above. The influencing factors such as the vehicle type and the load capacity may be input by a user, or may be obtained from vehicle information recorded in software preset in the vehicle at the time of shipment, but is not limited thereto.

The actual travel time (time of use) of the vehicle may be determined from the departure time and arrival time using timestamps collected by the vehicle terminals in conjunction with Global Positioning System (GPS) positioning. However, this is merely an example, and the actual travel time may be acquired by various known methods.

In addition, the historical driving information record can also record the specific driving track of the vehicle so as to judge whether various vehicles driving between the starting point and the target point travel the same route. The starting point, the target point and the driving track can be obtained by a global positioning system receiver mounted on the vehicle and sent to the historical driving information recording library to be recorded in the historical driving information record, but the invention is not limited thereto.

The structure of the prediction model creation unit 20 is explained in detail below with reference to fig. 2.

Fig. 2 is a functional block diagram showing the prediction model creating unit 20 of the device 1 for predicting the vehicle travel time according to the first embodiment of the present invention.

The prediction model creation unit 20 includes a collection unit 201, a first reference historical travel time calculation unit 203, an arrival state determination unit 205, a main influence factor determination unit 207, a grouping unit 209, and a second reference historical travel time calculation unit 211.

The collection unit 201 collects, from the historical travel information record base, historical travel information records including historical travel time and travel affecting factors from a start point to a target point, from the start point and the target point specified. Specifically, the collection unit 201 may collect n pieces of history travel information records from the start point to the target point, the history travel information records including history travel time from the start point to the target point and k kinds of travel influence factors (where n, k are natural numbers), according to the designated start point and target point. For example, the driving influence factors include 5 kinds of driving influence factors such as "departure time", "whether or not a holiday is public, a" weather condition "," a vehicle type "," a load ", and the like.

The first reference historical travel time calculation unit 203 calculates the reference historical travel time of the historical travel information record using the historical travel time collected by the collection unit 201. Specifically, the first reference historical travel time calculation unit 203 may calculate the reference historical travel time for the n historical travel information records as a whole from the historical travel time in each of the n historical travel information records. For example, the first reference historical travel time calculation unit 203 sets the median of all the historical travel times of the n historical travel information records as the reference historical travel time of the n historical travel information records as a whole.

The arrival state determination unit 205 determines the arrival state of the history travel information record by comparing the history travel time with the reference history travel time calculated by the first reference history travel time calculation unit 203. The arrival state indicates whether the historical travel time is late compared to the reference historical travel time, for example, indicates whether the historical travel time of each historical travel information record is "early arrival", or "on time arrival", or "late arrival" compared to the reference historical travel time. Specifically, the arrival state determination unit 205 compares the historical travel time of each of the n historical travel information records with the reference historical travel time of the n historical travel information records as a whole, respectively, determines the arrival state of each of the n historical travel information records, the arrival state indicating whether the historical travel time of each of the n historical travel information records is late compared with the reference historical travel time of the n historical travel information records as a whole, and the state value of the arrival state includes m arrival state values, where m is a natural number and 0 < m ≦ n. The state values of the arrival state may be specifically set according to actual needs, and for example, the state values of the arrival state may include three state values of "early arrival", "on-time arrival", and "late arrival" (m is 3), and for example, 5 state values (m is 5) such as "early arrival at 1 hour or more", "early arrival at 0.5 to 1 hour", "early arrival at 0 to 0.5 hour or less", "late arrival at 0.5 hour or more" may be included.

The main influence factor determination unit 207 analyzes the history travel information record based on the arrival state determined by the arrival state determination unit 205, thereby determining at least one main influence factor having the greatest influence on the arrival state among the travel influence factors. Specifically, the main influence factor determination unit 207 may analyze the n historical travel information records as a whole using a decision tree algorithm based on the determined arrival state of each of the n historical travel information records, thereby determining the main influence factor having the largest influence on the state value of the arrival state among the k travel influence factors, and divide the state value of the main influence factor into q state categories corresponding to m arrival state values, where q is a natural number and 0 < q ≦ m. The arrival state values and the state classes may be in one-to-one correspondence, where m is q, or multiple arrival state values may correspond to one state class, where q is less than m. For example, the main influence factor determination unit 207 determines the "departure time" as the main influence factor having the largest influence on the change of the arrival state value from among 5 kinds of travel influence factors such as "departure time", "whether or not on a public holiday", "weather conditions", "vehicle type", "load", and the like, after analyzing the entire history travel information record based on the arrival state including the three state values of "early arrival", "on-time arrival", and "late arrival", by using the decision tree algorithm, and determines the boundary point at which the main influence factor affects the arrival state value, for example, when the "departure time" is earlier than "6 am", the arrival state values are all "early arrival", when the "departure time" is "6 am", the arrival state value is "late arrival" when the "departure time" is later than "6 am", thus, it is determined that the state value "6 o 'clock" of the main influence factor is a boundary point that influences the arrival state value, and the state values of "departure time" as the main influence factor are classified into three state categories of "earlier than 6 o' clock", "6 o 'clock", and "later than 6 o' clock" (q is m is 3) in correspondence with three arrival state values (m is 3) of "early arrival", "just arrival", and "late arrival". Further, the "6 o 'clock" and the "earlier than 6 o' clock" as the boundary point may be combined, and the status value of the "departure time" may be divided into two status categories of "no later than 6 o 'clock" and "later than 6 o' clock" (q ═ 2 < m). Here, the primary influencing factor is shown as one, but this is merely an example, and in some embodiments, the primary influencing factor may be a plurality of the travel influencing factors.

Further, the grouping unit 209 groups the history travel information records according to the main influence factor determined by the main influence factor determination unit 207. Specifically, the grouping unit 209 may divide the n history traveling information records into corresponding q groups of history traveling information records according to the state classification of the main influence factor of the n history traveling information records. For example, grouping section 209 classifies history travel information records into three sets of history travel information records (q-m-3) in which the state value of "departure time" is earlier than 6 points, the state value of "departure time" is 6 points, and the state value of "departure time" is later than 6 points, in accordance with three state categories (q-m-3) of "earlier than 6 points", "6 points", and "later than 6 points" of "departure time", which are main influence factors. Alternatively, the grouping unit 209 may correspondingly group the historical travel information records into two sets of historical travel information records (q 2 < m) having a status value of "departure time" of not later than 6 o ' clock "and a status value of" departure time "of not later than 6 o ' clock", based on two status categories (q 2 < m) of "departure time" and "later than 6 o ' clock" as main influencing factors.

The second reference historical travel time calculation unit 211 calculates the reference historical travel time of each set of historical travel information records grouped by the grouping unit 209. Specifically, the second reference historical travel time calculation unit may calculate the reference historical travel time for each of the historical travel information records of each of the q sets of historical travel information records as a whole, from the historical travel time for each of the historical travel information records of each of the q sets of historical travel information records. For example, the second reference historical travel time calculation unit 211 takes the median of all the historical travel times in each set of historical travel information records as the reference historical travel time for the entire set of historical travel information records.

The prediction model is built by the cooperative operation of the above-described collection unit 201, first reference historical travel time calculation unit 203, arrival state determination unit 205, main influence factor determination unit 207, grouping unit 209, and second reference historical travel time calculation unit 211.

Returning to fig. 1, the description of vehicle travel time prediction section 30 is continued.

The vehicle travel time prediction unit 30 predicts the vehicle travel time from the start point to the target point using the above prediction model that has been established, based on the current actual input conditions including the start point as the vehicle departure point, the target point as the vehicle destination, and the main influence factor, whenever it is necessary to predict the vehicle travel time from the start point to the target point. Specifically, the vehicle travel time prediction unit 30 may determine a set of historical travel information records corresponding to the state category of the main influence factor in the current actual input condition according to the current actual input condition, and predict the reference historical travel time of the set of historical travel information records as the vehicle travel time. For example, when the user of the vehicle inputs the current actual input conditions of the state value "6 o 'clock" of the "departure time", the start point "a ground", and the target point "B ground" as the main influence factors, the set of historical travel information records in which the state value of the "departure time" is "6 o' clock" is determined using the established prediction model, and the reference historical travel time of the set of historical travel information records is predicted as the travel time of the vehicle.

Next, a method of predicting the actual vehicle travel according to the first embodiment of the present invention will be described in detail.

As shown in fig. 3, the method of predicting the vehicle travel time of the first embodiment of the present invention includes a prediction model creation step S1 and a vehicle travel time prediction step S2.

In the prediction model creation step S1, the prediction model creation unit 20 collects historical travel information records from a start point to a target point from a historical travel information record library on the basis of the designated start point and target point, calculates a reference historical travel time of the historical travel information records using the historical travel information records, and determines a main influence factor having the greatest influence on the travel time among the travel influence factors to create a prediction model.

Next, the prediction model creation step S1 will be described in detail.

As shown in fig. 4, the above-described prediction model establishing step S1 may include a collecting step S201, a first reference historical travel time calculating step S203, an arrival state determining step S205, a main influence factor determining step S207, a grouping step S209, and a second reference historical travel time calculating step S211.

Specifically, in the collection step S201, the collection unit 201 collects, from the history travel information record base, the history travel information records from the start point to the target point, including the history travel time from the start point to the target point and the travel influence factor, according to the designated start point and target point. For example, in the collecting step S201, n pieces of history travel information records from the start point to the target point, which include the history travel time from the start point to the target point and k kinds of travel influence factors, where n, k are natural numbers, are collected according to the designated start point and target point.

Next, in the first reference historical travel time calculation step S203, the first reference historical travel time calculation unit 203 calculates the reference historical travel time of the historical travel information record using the historical travel time of the historical travel information record collected by the collection unit 201 in the collection step S201. For example, in the first reference historical travel time calculation step S203, the reference historical travel time of the entire n pieces of historical travel information records is calculated from the historical travel time in each of the n pieces of historical travel information records. Specifically, in the first reference historical travel time calculating step S203, the median of all the historical travel times of the n pieces of historical travel information records may be set as the reference historical travel time of the n pieces of historical travel information records as a whole.

After the reference historical travel time of the history travel information record as a whole is determined in the first reference historical travel time calculation step S203, the arrival state determination unit 205 determines the arrival state of the history travel information record indicating whether the history travel time is late compared with the reference historical travel time by comparing the history travel time with the reference historical travel time in the arrival state determination step S205. For example, in the arrival state determining step, the historical travel time of each of the n historical travel information records is compared with the reference historical travel time of the n historical travel information records as a whole, respectively, and the arrival state of each of the n historical travel information records is determined, the arrival state indicating whether the historical travel time of each of the n historical travel information records is later than the reference historical travel time of the n historical travel information records as a whole, and the state value of the arrival state includes m arrival state values, where m is a natural number and 0 < m ≦ n.

After determining the arrival state in the arrival state determining step S205, in the main influence factor determining step S207, the main influence factor determining unit 207 analyzes the history travel information record based on the arrival state determined by the arrival state determining unit 205 in the arrival state determining step S205, thereby determining at least one main influence factor having the greatest influence on the arrival state among the travel influence factors. For example, in the main influence factor determination step S207, the n historical travel information records are analyzed as a whole using a decision tree algorithm based on the arrival state of each of the n determined historical travel information records, thereby determining the main influence factor having the largest influence on the state value of the arrival state among the k travel influence factors, and dividing the state value of the main influence factor into q state categories corresponding to m arrival state values, where q is a natural number and 0 < q ≦ m.

Next, in the grouping step S209, the grouping unit 209 groups the history travel information records according to the main influence determined in the main influence determining step S207. Specifically, in the grouping step S209, the n pieces of historical travel information are divided into corresponding q groups of historical travel information according to the state categories of the main influence factors of the n pieces of historical travel information.

After the grouping is completed in the grouping step S209, the second reference historical travel time calculation unit 211 calculates the reference historical travel time for each set of the historical travel information records after the grouping in a second reference historical travel time calculation step S211. Specifically, in the second reference historical travel time calculating step S211, the reference historical travel time of the entire set of historical travel information records is calculated from the historical travel time of each historical travel information record in each of the q sets of historical travel information records. For example, in the second reference historical travel time calculation step S211, the median of all the historical travel times in each set of historical travel information records is set as the reference historical travel time for the entire set of historical travel information records.

The establishment of the prediction model is completed through the steps S201 to S211 as described above.

Next, returning to fig. 3, when the user needs to predict the vehicle travel time from a start point to a target point, in a vehicle travel time prediction step S2, the vehicle travel time prediction unit 30 predicts the vehicle travel time from the start point to the target point using the already-established prediction model according to current actual input conditions including the start point as the vehicle departure point, the target point as the vehicle destination, and the main influence factor. Specifically, in the vehicle travel time prediction step S2, a set of historical travel information records corresponding to the status category of the primary influence factor in the current actual input condition is determined based on the current actual input condition, and the reference historical travel time of the set of historical travel information records is predicted as the vehicle travel time.

After the prediction model creation unit 20 creates the prediction model in the prediction model creation step S1 described above, the vehicle travel time prediction unit can accurately predict the travel time of the vehicle from the start point to the target point in the vehicle travel time prediction step S2 using the prediction model created in the prediction model creation step S1.

According to the system and method for predicting vehicle travel time of the first embodiment of the present invention as described above, since the influence of the abnormal value is excluded by using the reference historical travel time, the vehicle travel time can be predicted with high accuracy. In addition, the prediction model established offline according to the historical driving information record is adopted, and when the user inputs the current actual driving condition, the driving time of the vehicle is predicted by using the prediction model, so that the data volume to be processed by the processor in real time is greatly reduced, the load of the processor is reduced, the prediction result can be obtained quickly, and the prediction efficiency is improved. The method and the system can be used for short-distance prediction such as prediction of vehicle running time in cities and long-distance prediction such as long-distance logistics distribution.

In addition, the historical travel information records are analyzed by using a decision tree algorithm, main influence factors influencing the travel time of the vehicle are determined, the historical travel information records are grouped according to the main influence factors, and the reference historical travel time of a group of historical travel information records corresponding to the state type of the main influence factors in the current actual input condition is predicted as the travel time of the vehicle. Therefore, the accuracy of predicting the vehicle travel time is improved, and when the user inputs the current actual travel condition, the prediction result can be determined by quickly finding the reference historical travel time of the corresponding set of historical travel information records without complicated calculation, thereby improving the prediction efficiency.

In addition, since the median of the historical travel time is used as the reference historical travel time, the influence of the abnormal value in prediction such as the average value on the prediction result is eliminated, and the accuracy of predicting the travel time is further improved.

[ second embodiment ]

Hereinafter, an apparatus and method for predicting a vehicle travel time according to a second embodiment of the present invention will be described in detail with reference to fig. 5 and 6.

Fig. 5 is a block diagram showing a device 11 for predicting a vehicle travel time according to a second embodiment of the present invention.

As shown in fig. 5, the apparatus 11 for predicting the traveling practice of a vehicle according to the second embodiment of the present invention includes a prediction model creation unit 120, a vehicle traveling time prediction unit 130, and a self-learning unit 140. The configurations of the prediction model creation unit 120 and the vehicle travel time prediction unit 130 are the same as those of the prediction model creation unit 20 and the vehicle travel time prediction unit 30 in the first embodiment described above, respectively, and therefore detailed description thereof is omitted. Here, only the self-learning unit 140 will be described in detail.

The self-learning unit 140 writes the actual travel time of the vehicle from the start point to the target point in association with the start point, the target point, and the main influence factor in the history travel information record to update the prediction model after the vehicle reaches the destination after the vehicle travel time prediction is completed as described above. The self-learning unit 140 may immediately associate the actual travel time of the vehicle with the start point, the target point, and the main influence factor and write the associated information into the history travel information record base, and immediately repeat the prediction model building step to update the prediction model each time the vehicle reaches the destination. Alternatively, the self-learning unit 140 may associate the actual travel time of the vehicle with the start point, the target point, and the main influence factor and write the associated data into the memory every time the vehicle reaches the destination, and after a certain amount of data is accumulated, write the data of the actual travel time accumulated in the memory into the historical travel information record every predetermined period (for example, one month) to update the prediction model. The latter is preferred for stability of the predictive model and good user experience.

As shown in fig. 6, the method of predicting the traveling practice of the vehicle of the second embodiment of the present invention includes a prediction model establishing step S11, a vehicle traveling time predicting step S12, and a self-learning step S13. The prediction model establishing step S11 and the vehicle travel time predicting step S12 are similar to the prediction model establishing step S1 and the vehicle travel time predicting step S2 in the above-described first embodiment, respectively, and thus detailed description thereof is omitted. Here, only the self-learning step S13 will be described.

In the self-learning step S13, each time the vehicle reaches the destination, the self-learning unit 140 writes the actual travel time of the vehicle from the start point to the target point in association with the start point, the target point, and the main influence factor in the history travel information record to update the prediction model, thereby realizing self-learning. In the self-learning step S13, the self-learning unit 140 may write the actual travel time of the vehicle in association with the start point, the target point, and the main influence factor into the historical travel information record base immediately after the vehicle reaches the destination, and immediately repeat the prediction model establishing step to update the prediction model. Alternatively, after the vehicle reaches the destination each time, the actual travel time of the vehicle may be associated with the start point, the target point, and the main influence factor and written into the memory, and after a certain amount of the actual travel time is accumulated, the actual travel time data record accumulated in the memory may be written into the history travel information record at predetermined intervals (for example, one month) to update the prediction model. The latter is preferred for stability of the predictive model and good user experience.

The apparatus and method for predicting a vehicle travel time according to the second embodiment of the present invention can further improve the accuracy of predicting a vehicle travel time by updating the history travel information record with the newly generated travel information record at regular intervals through the self-learning step by the self-learning unit, in addition to the effects achieved in the first embodiment.

[ DEFORMATION ] OF THE PREFERRED EMBODIMENT

In the first and second embodiments described above, the history travel information records are grouped once according to the status type of the main influence factor. However, it is also possible to group each set of the history traveling information records after the primary grouping again based on the state type of the next traveling influencing factor, which mainly influences the arrival state, in addition to the main influencing factor, and calculate the reference history traveling time of each set of the history traveling information records after the corresponding secondary grouping. Such modifications also fall within the scope of the present invention. At this time, when predicting the vehicle travel time, the vehicle travel time prediction unit 30 determines a set of historical travel information records corresponding to the state type of the main influence factor and the next travel influence factor that mainly affects the arrival state in the current actual input condition, and then predicts the reference historical travel time of the set of historical travel information records as the travel time of the vehicle.

[ examples ] A method for producing a compound

Next, an example of predicting the vehicle travel time using the apparatus and method for predicting the vehicle travel time according to the first or second embodiment of the present invention will be described with reference to fig. 7 to 10.

Fig. 7 is an example table showing a vehicle history travel information record of the embodiment of the invention. Fig. 8 is an exemplary table showing a vehicle history traveling information record after determining the arrival state of the vehicle according to the embodiment of the present invention. FIG. 9 is an example of a process for determining primary influencing factors using a decision tree algorithm that represents an embodiment of the present invention. Fig. 10 is an exemplary table showing the grouping of historical travel information records after determining a main influence factor using a decision tree algorithm according to an embodiment of the present invention.

First, before predicting the vehicle travel time, the prediction model creation unit 20 or 120 creates a prediction model.

Specifically, in the first step, the collection unit 201 of the prediction model creation unit 20 or 120 collects, for example, history travel information records on the same route between a specified start point and a target point, for example, from the city a to the city B, including travel influence factors such as departure date, departure time, vehicle type, load, weather condition, whether or not a public holiday and history travel time, from the history travel information record base, based on the GPS track on which the vehicle travels and public information at that time. For example, as shown in fig. 7, the collection unit 201 collects 15 history travel information records from point a to point B. The GPS track is mainly used for judging whether the vehicles run on the same route or not.

Then, at the second step, the first reference historical travel time calculation unit 203 calculates the reference historical travel time on the same route from the historical travel information collected at the first step, for the same start point and target point. At this time, the median of all the historical travel times is adopted as the reference historical travel time. For example, as shown in fig. 7, the reference historical travel time of a to B on this line is calculated to be 7 hours using the median of 15 historical travel times.

Next, at the third step, the arrival state determination unit 205 compares each of the historical travel times with the reference historical travel time calculated at the second step for each of the historical travel information records, and determines the arrival state. For example, as shown in fig. 8, if the travel time is less than or equal to the reference history travel time, it is determined that the state value of the arrival state is "not late"; if it is greater than the reference history travel time, the state value of the arrival state is determined to be "late".

Then, in the fourth step, the main influence factor determination unit 207 determines the main influence factor having the largest influence on the arrival state by using a decision tree algorithm. For example, as shown in fig. 9, using a decision tree algorithm, it is determined that "departure time" is the main influence factor from a decision tree obtained from the data in fig. 8, and it is determined that the demarcation point that influences the arrival state value is 6 am, and the state value of the departure time as the main influence factor is divided into two state categories of "not later than 6 am" and "later than 6 am".

Next, at the fifth step, the grouping unit 209 groups the history traveling information records according to the state type of the state value of the main influence factor obtained at the fourth step. For example, as shown in fig. 10, the first 8 records are divided into a first group (departure time no later than 6 o 'clock, indicated by dark shading in the figure), and the last 7 records are divided into a second group (departure time no later than 6 o' clock, indicated by light shading in the figure).

Next, at the sixth step, the second reference historical travel time calculation unit 211 calculates the reference historical travel time for each set of historical travel information records obtained at the fifth step, for each set of historical travel information records. In calculating the reference historical travel time, in order to exclude the influence of the abnormal value, the median of the historical travel urgency records of each group is used, and for example, for the first group in fig. 10, the median is used as the reference historical travel time of the first group, that is, 6.85 hours, and for the second group, the median is used as the reference historical travel time of the second group, that is, 8.5 hours.

Thereby, the prediction model creation unit 20 completes the creation of the prediction model.

After the prediction model is established, when it is necessary to predict the travel time of the vehicle from the a ground to the B ground, the vehicle travel time prediction unit 30 or 130 determines a group corresponding to the state class of the state value of the departure time as the main influence factor based on the actual shift period input conditions, such as the start point and the target point, the departure time, and the like, and takes the reference history travel time of the group as the predicted travel time. For example, if departure is made from point a to point B at 5 am, the reference historical travel time of the first set of historical travel information records having the departure time status category of "no later than 6 o" is predicted as the vehicle travel time, that is, it is predicted that 6.85 hours are required from a to B; in addition, if departure is made at 15 pm, the reference historical travel time recorded with the second group of historical travel information of which the departure time status category is "later than 6" is predicted as the vehicle travel time, that is, 8.5 hours is predicted to be required from a to B.

Thereby, the prediction of the travel time of the vehicle from point a to point B is completed.

In the embodiment as described above, after 15 pieces of history travel information records of the same route from point a to point B are calculated to have a reference history travel time of 7 hours, and the history travel information records are divided into two arrival states of "not late" and "late", the "departure time" is determined as a main influence factor that mainly influences the arrival state according to the decision tree algorithm, and the demarcation point at which the arrival state value changes is determined as "6 am", thereby dividing the state value of the main influence factor "departure time" into two state categories of "not later than 6 am" and "later than 6 am", thereafter, the above 15 pieces of history travel information are divided into two groups according to the two state categories, and the reference history travel time is calculated for each group using the median thereof, and after the prediction model is established, when the vehicle travel time is to be predicted, after a corresponding set of historical travel information records is determined based on actual input conditions, the reference historical travel time of the set is predicted as the vehicle travel time. This eliminates the influence of an abnormal value that deviates too much from the reference historical travel time in the historical travel information record, and for example, as shown in fig. 7, 8, and 10, the influence of an abnormal value that the historical travel time is 17 hours can be eliminated. Thereby improving the accuracy in predicting the travel time. Also, the historical travel information records are grouped according to the main influence factor that influences the travel time, so the accuracy of predicting the travel time is improved compared to predicting the vehicle travel time using the reference travel time of all data.

In addition, the offline prediction model established according to the historical driving information record is adopted, and when the user inputs the current actual driving condition, the driving time of the vehicle is rapidly predicted by using the offline prediction model, so that the data volume to be processed by the processor in real time is greatly reduced, the load of the processor is reduced, the prediction result can be rapidly obtained, and the prediction efficiency is improved.

In addition, since the historical travel information can be grouped in the historical travel information record according to the travel influence factors such as the vehicle type and the vehicle application, the present invention can be used for short-distance prediction such as prediction of vehicle travel time in cities, and long-distance prediction such as long-distance logistics distribution.

In addition, in the embodiment, the decision tree algorithm is adopted, so that the algorithm is simple, quick and effective on the basis of ensuring the calculation accuracy, and the prediction efficiency is improved.

In the above embodiment, although not mentioned, when the device 11 for predicting the vehicle travel time according to the second embodiment is used for prediction, the prediction result may be self-learned, and therefore the prediction accuracy in the future can be further improved.

The various techniques, embodiments, examples, or certain aspects or portions thereof, units (e.g., the historical travel information log base, the collection unit 201 of the prediction model creation unit 20, the first reference historical travel time calculation unit 203, the arrival state determination unit 205, the primary influence factor determination unit 207, the grouping unit 209, and the second reference historical travel time calculation unit 211, etc.) described above may take the form of program code (i.e., instructions) embodied in tangible media, such as various circuitry, a floppy disk, a compact disc-read only memory (CD-ROM), a hard disk drive, a non-transitory computer-readable storage medium, or any other machine-readable storage medium, wherein, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the various techniques. The circuitry may include hardware, firmware, program code, executable code, computer instructions, and/or software. The non-transitory computer-readable storage medium may be a computer-readable storage medium that does not include a signal. In the case of program code execution on programmable computers, the computer may include a processor, a storage medium readable by the processor (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. The volatile and non-volatile memory and/or storage elements may be Random Access Memory (RAM), Erasable Programmable Read Only Memory (EPROM), flash drives, optical disk drives, hard disk drives, solid state drives, or other media for storing electronic data. One or more programs that may implement or utilize the various techniques described herein may use an Application Programming Interface (API), multiplexing controls, and the like. However, the program or programs may be implemented in component or machine language, if desired. In any case, the language may be an assembly or interpreted language, and combined with hardware implementations.

It should be understood that many functional units described in this specification have been labeled as units (e.g., the historical travel information log base, the collection unit 201 of the prediction model creation unit 20, the first reference historical travel time calculation unit 203, the arrival state determination unit 205, the main influence factor determination unit 207, the grouping unit 209, and the second reference historical travel time calculation unit 211, etc.) in order to more particularly emphasize their implementation independence. For example, these elements may be implemented as hardware circuits comprising custom very-large-scale integration (VLSI) circuits or gate arrays, off-the-shelf semiconductors such as logic chips, transistors, or other discrete components. A cell may also be implemented in programmable hardware devices such as field programmable gate arrays, programmable logic devices or the like.

These units may also be implemented in software for execution by various types of processors. An identified unit of executable code may, for instance, comprise one or more physical or logical blocks of computer instructions which may, for instance, be organized as an object, procedure, or function. Nevertheless, the executables of an identified unit need not be physically located together, but may comprise discrete instructions stored in different locations which, when joined logically together, comprise the unit and achieve the stated purpose for the unit.

Of course, a unit of executable code may be a single instruction, or many instructions, and may even be distributed over several different code segments, among different programs, and across several memory devices. Similarly, operational data may be identified and illustrated herein within elements, and may be embodied in any suitable form and organized within any suitable type of data structure. The operational data may be collected as a single data set, or may be distributed over different locations including over different storage devices, and may exist, at least partially, merely as electronic signals on a system or network. The cells may be passive or active, including agents operable to perform desired functions.

The respective units of the collection unit 201, the first reference historical travel time calculation unit 203, the arrival state determination unit 205, the main influence factor determination unit 207, the grouping unit 209, the second reference historical travel time calculation unit 211, and the like of the present invention described above may be implemented by machines. In alternative embodiments, the machine may operate as a standalone device or may be connected (e.g., networked) to other machines. In a networked deployment, the machine may operate in the capacity of a server machine and/or a client machine in server-client network environment. The machine may be a vehicle terminal, an AVN (audio video navigation), a Personal Computer (PC), a tablet PC, a set-top box (STB), a Personal Digital Assistant (PDA), a mobile telephone, a network appliance, a network router, switch or bridge, or any machine capable of executing instructions (sequential or otherwise) that specify actions to be taken by that machine. Further, the term "machine" can also include any collection of machines that individually or jointly execute a set (or multiple sets) of instructions to implement any one or more of the units or methods discussed herein, such as cloud computing, software as a service (SaaS), other computer cluster configurations, and so forth. For example, the prediction model building unit 20 may be implemented in a cloud server, and the vehicle travel time prediction unit 30 may be implemented in a navigation system of a vehicle terminal, and jointly implement the apparatus for predicting vehicle travel time of the present invention through a network, or jointly execute the method for predicting vehicle travel time of the present invention.

Machines (e.g., cloud servers, in-vehicle terminals, computer systems) may include a hardware processor (e.g., a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a hardware processor core, or any combination thereof), a main memory, and a static memory, at least some of which may communicate with others via a link (e.g., a bus). The machine may also include a display unit, an alphanumeric input device (e.g., a keyboard), and a User Interface (UI) navigation device (e.g., a mouse). In an example, the display unit, the input device, and the UI navigation device may be a touch screen display. The machine may additionally include a storage device (e.g., a drive unit), a signal generation device (e.g., a speaker), a network interface device, and one or more sensors, such as a Global Positioning System (GPS) sensor, compass, accelerometer, or other sensor. The machine may include an output controller 728 such as a serial (e.g., Universal Serial Bus (USB), parallel, or other wired or wireless (e.g., Infrared (IR)) connection to communicate or control one or more peripheral devices (e.g., a printer, card reader, etc.).

The storage device may include at least one machine-readable medium having stored thereon one or more sets of data structures or instructions (e.g., software) embodied in or utilized by any one or more of the techniques or functions described herein. The instructions may also reside, at least partially, on additional machine-readable memory, such as main memory, static memory, or within a hardware processor during execution thereof by the machine. In an example, one or any combination of a hardware processor, a main memory, a static memory, or a storage device may constitute a machine-readable medium.

The term "machine-readable medium" can include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) configured to store the one or more instructions.

The term "machine-readable medium" can include any medium that is capable of storing, encoding or executing instructions for execution by the machine and that cause the machine to perform any one or more of the techniques of this disclosure, or that is capable of storing, encoding or executing data structures used by or associated with such instructions. Examples of machine-readable media may include, but are not limited to, solid-state memories, and optical and magnetic media. Specific examples of the machine-readable medium may include: nonvolatile memories such as semiconductor Memory devices (e.g., Electrically Programmable Read-Only memories (EPROMs), Electrically Erasable Programmable Read-Only memories (EEPROMs), and flash Memory devices); magnetic disks such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks.

The instructions may further be transmitted or received over a communication network using a transmission medium via a network interface device utilizing any of a number of transfer protocols (e.g., frame relay, Internet Protocol (IP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), hypertext transfer protocol (HTTP), etc.). Example communication networks may include Local Area Networks (LANs), Wide Area Networks (WANs), packet data networks (e.g., the Internet), Mobile Telephone networks (e.g., channel Access methods including Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), and Orthogonal Frequency Division Multiple Access (OFDMA), as well as cellular networks, such as Global System for Mobile Communications (GSM), Universal Mobile Communications (UMTS), CDMA 20001 x standards and Long Term Evolution (LTE), legacy Mobile Telecommunications (POTS), and IEEE, e.g., Institute of Electrical and Electronic Engineers (IEEE), including IEEE 802.11 Standard (WiFi), IEEE802.16 standard

And others), a point-to-point (P2P) network, or a CAN-like vehicle network or other protocol now known or later developed.

Further, it should be understood that the term "vehicle" or other similar term as used herein includes motor vehicles in general, such as passenger vehicles including Sport Utility Vehicles (SUVs), buses, trucks, various commercial vehicles, motorcycles, and the like, and includes hybrid electric vehicles, electric-only vehicles, plug-in hybrid electric vehicles, hydrogen-powered vehicles, and other alternative fuel vehicles (e.g., fuels derived from resources other than petroleum).

Although the preferred embodiments of the present invention have been described for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims.

Claims

1. A method of predicting a travel time of a vehicle, comprising:

a prediction model establishing step of collecting historical travel information records from a starting point to a target point from a historical travel information record base according to a specified starting point and target point, calculating reference historical travel time of the historical travel information records by using the historical travel information records, and determining at least one main influence factor having the greatest influence on the travel time among the travel influence factors to establish a prediction model; and

a vehicle travel time prediction step of, when it is necessary to predict a vehicle travel time from the start point to the target point, predicting the vehicle travel time from the start point to the target point using the prediction model that has been established, based on current actual input conditions including the start point as a vehicle departure place, the target point as a vehicle destination, and the main influence factor,

the step of establishing the prediction model comprises the following steps:

a collection step of collecting a history travel information record from a start point to a target point from the history travel information record base according to a specified start point and target point, the history travel information record including history travel time and travel influence factors from the start point to the target point;

a first reference historical travel time calculation step of calculating a reference historical travel time of the historical travel information record using the historical travel time,

an arrival state determination step of determining an arrival state of the history travel information record by comparing the history travel time with the reference history travel time, the arrival state indicating whether the history travel time is late compared with the reference history travel time;

a main influence factor determination step of analyzing the historical travel information record based on the arrival state, thereby determining at least one main influence factor having the greatest influence on the arrival state among the travel influence factors;

grouping, namely grouping the historical driving information records according to the main influence factors; and

and a second reference historical travel time calculation step of calculating the reference historical travel time of each group of the grouped historical travel information records.

2. The method of predicting a travel time of a vehicle according to claim 1, characterized in that:

in the collecting step, n historical travel information records from the starting point to the target point are collected according to the specified starting point and the specified target point, the historical travel information records comprise historical travel time from the starting point to the target point and k travel influence factors, wherein n and k are natural numbers;

in the first reference historical travel time calculation step, a reference historical travel time of the n historical travel information records as a whole is calculated from the historical travel time in each of the n historical travel information records;

in the arrival state determining step, comparing the historical travel time of each of the n historical travel information records with the reference historical travel time of the n historical travel information records as a whole, respectively, to determine the arrival state of each of the n historical travel information records, the arrival state indicating whether the historical travel time of each of the n historical travel information records is late compared with the reference historical travel time of the n historical travel information records as a whole, and the state values of the arrival state including m arrival state values, where m is a natural number and 0 < m ≦ n;

in the main influence factor determination step, based on the determined arrival state of each of the n historical travel information records, analyzing the n historical travel information records as a whole by using a decision tree algorithm, thereby determining a main influence factor having the largest influence on the state value of the arrival state among the k travel influence factors, and dividing the state value of the main influence factor into m state categories corresponding to the m arrival state values;

in the grouping step, the n historical travel information records are divided into m groups of corresponding historical travel information records according to the state type of the main influence factor of the n historical travel information records; and

in the second reference historical travel time calculation step, the reference historical travel time for the entire set of historical travel information records is calculated from the historical travel time for each of the historical travel information records in each of the m sets of historical travel information records.

3. The method of predicting a travel time of a vehicle according to claim 2, characterized in that:

in the vehicle travel time prediction step, a set of historical travel information records corresponding to the state category of the main influence factor in the current actual input condition is determined according to the current actual input condition, and a reference historical travel time of the set of historical travel information records is predicted as the vehicle travel time.

4. A method of predicting a travel time of a vehicle according to claim 2 or 3, characterized in that:

in the first reference historical travel time calculation step, a median of all the historical travel times of the n historical travel information records is set as the reference historical travel time of the entire n historical travel information records.

5. A method of predicting a travel time of a vehicle according to claim 2 or 3, characterized in that:

in the second reference historical travel time calculation step, a median of all the historical travel times in each set of historical travel information records is set as the reference historical travel time for the entire set of historical travel information records.

6. The method of predicting a travel time of a vehicle according to claim 1 or 2, further comprising:

and a self-learning step of writing the actual travel time of the vehicle from the starting point to the target point into the historical travel information record in association with the starting point, the target point and the main influence factor each time the vehicle reaches the destination, so as to update the prediction model.

7. An apparatus for predicting a travel time of a vehicle, comprising:

a prediction model establishing unit which collects historical driving information records from a starting point to a target point from a historical driving information record base according to a specified starting point and the target point, calculates reference historical driving time of the historical driving information records by using the historical driving information records, and determines at least one main influence factor which has the largest influence on the driving time in the driving influence factors to establish a prediction model;

a vehicle travel time prediction unit that predicts a vehicle travel time from the start point to the target point using the prediction model that has been established, when it is necessary to predict a vehicle travel time from the start point to the target point, based on current actual input conditions including the start point as a vehicle departure place, the target point as a vehicle destination, and the main influence factor,

the prediction model building unit includes:

a collection unit that collects, from the historical travel information record base, historical travel information records from a start point to a target point, the historical travel information records including historical travel time and travel influence factors from the start point to the target point, according to a specified start point and target point;

a first reference historical travel time calculation unit that calculates a reference historical travel time of the historical travel information record using the historical travel time,

an arrival state determination unit that determines an arrival state of the history travel information record by comparing the history travel time with the reference history travel time, the arrival state indicating whether the history travel time is late compared with the reference history travel time;

a main influence factor determination unit that analyzes the historical travel information record based on the arrival state, thereby determining at least one main influence factor that has the greatest influence on the arrival state among the travel influence factors;

a grouping unit that groups the history travel information records according to the main influence factor; and

and a second reference historical travel time calculation unit which calculates the reference historical travel time of each group of the grouped historical travel information records.

8. The apparatus for predicting the travel time of a vehicle according to claim 7, characterized in that:

the collecting unit collects n historical driving information records from the starting point to the target point according to the specified starting point and the specified target point, wherein the historical driving information records comprise historical driving time from the starting point to the target point and k driving influence factors, and n and k are natural numbers;

the first reference historical travel time calculation unit calculates a reference historical travel time for the n historical travel information records as a whole, based on the historical travel time in each of the n historical travel information records;

the arrival state determination unit compares the historical travel time of each of the n historical travel information records with a reference historical travel time of the n historical travel information records in their entirety, respectively, and determines an arrival state of each of the n historical travel information records indicating whether the historical travel time of each of the n historical travel information records is late compared with the reference historical travel time of the n historical travel information records in their entirety, and the state values of the arrival state include m arrival state values, where m is a natural number and 0 < m ≦ n;

the main influence factor determination unit analyzes the n historical driving information records as a whole by using a decision tree algorithm based on the determined arrival state of each historical driving information record in the n historical driving information records, thereby determining a main influence factor having the largest influence on the state value of the arrival state among the k driving influence factors, and dividing the state value of the main influence factor into m state categories corresponding to the m arrival state values;

the grouping unit divides the n historical driving information records into m groups of corresponding historical driving information records according to the state types of main influence factors of the n historical driving information records; and

the second reference historical travel time calculation unit calculates the reference historical travel time for each of the sets of historical travel information records as a whole, based on the historical travel time for each of the sets of historical travel information records in each of the m sets of historical travel information records.