CN106251642B - A kind of public transport road chain speed calculation method based on real-time bus GPS data - Google Patents

Abstract

The invention discloses a bus link speed calculation method based on real-time bus GPS data, which belongs to the intelligent transportation technology, and includes two parts: speed sample acquisition and link speed synthesis. Among them, the speed sample acquisition part maintains the GPS sampling point queue to generate a key sampling point sequence, and then calculates the speed sample; the link speed synthesis part maintains the speed sample queue of the road link, determines the speed sample size for calculation, and then synthesizes Get the link speed. The bus link speed calculation method provided by the present invention eliminates the influence of bus stops on the link speed calculation within the station range by generating a sequence of key sampling points and then calculating the speed samples, and improves the accuracy of the bus link speed Spend. The invention dynamically determines the sample size used for calculation according to the state of the speed sample queue, and performs dynamic calculation according to the real-time changing traffic conditions, and has better real-time performance and accuracy.

Description

A calculation method of bus link speed based on real-time bus GPS data

technical field

The invention belongs to the technical field of intelligent transportation, and in particular relates to a bus link speed calculation method based on real-time bus GPS data.

Background technique

With the continuous deepening of urban modernization, the number of motor vehicles is increasing rapidly, and the travel needs of citizens are diversified and personalized, and the pressure on urban traffic is increasing day by day. The development experience of large cities at home and abroad shows that to solve urban traffic problems, especially in super-large international cities, it is necessary to give full play to the important role of public transport. Facts have proved that the intelligent transportation system is an effective means to alleviate urban traffic congestion and solve the traffic dilemma in big cities. At present, the research hotspots of intelligent transportation systems, such as real-time traffic conditions and real-time bus arrival time prediction services, all require real-time calculation results of bus link speeds to support the above-mentioned functional services. However, the traffic situation is complex, changeable and unstable. How to obtain the driving speed of the bus network link in real time and accurately has become a key point of the intelligent bus system. At present, many studies have been carried out on road calculation at home and abroad, mainly including multiple regression estimation method, time series model, Kalman filter method and similarity model based on historical data.

The multiple regression estimation method is a method of predicting by analyzing the causal relationship and the degree of influence between things, such as analyzing the correlation between road speed and line structure, the number of traffic light intersections, the number of intermediate stations, weather and other factors, and constructing a regression equation , the dependent variable bus running time is predicted based on changes in these independent variables over the forecast period. This method requires these influencing factors to be independent, which strictly limits the application of regression analysis.

Time series forecasting is a method of inferring the future value of the research object by discovering the past change law of the research object. The accuracy of the prediction depends on the matching degree between the predicted travel time change law and the historical law, which has certain limitations.

Kalman filter is a high-efficiency recursive filter that can estimate the current position and future position from a set of observation sequences of object positions that contain noise. Kalman filtering is a method of predicting short-term traffic flow, which has good robustness and can respond effectively to the external environment. However, it is only effective in predicting the running time in one or two time periods in the future. When the predicted distance or overall time is very long, the prediction accuracy is low.

The prediction method based on the similarity model of historical data is based on the premise that the traffic operation has the regularity of cyclical changes. Through the analysis of historical data, the current link speed of the current bus network is predicted by the speed of the same link in the same historical period. On this basis, some studies have further proposed a road speed prediction mechanism based on road condition similarity, proposed road condition information tuples, and a method for calculating the similarity of road condition information tuples, and made predictions based on similar historical road conditions. This method is entirely based on historical data, does not take into account real-time changing traffic information, poor real-time performance and accuracy.

Contents of the invention

The technology of the present invention solves the problem: overcomes the impact of station delay on speed calculation, solves the problem of insufficient link coverage under different speed sample numbers, provides a bus link speed calculation method based on real-time bus GPS data, and solves the problem of bus link speed. The real-time requirement of link speed calculation improves the coverage and accuracy of the link, and provides data support for bus arrival forecast and real-time road condition expression of the bus network, which is robust and universal.

The invention preprocesses the real-time bus GPS point data collected in real time and matched to the bus line sampling point, then classifies and maintains the GPS sampling point queue according to the vehicle, extracts key points and calculates the speed sample for eliminating the influence of the station; adopts a dynamic window The mechanism maintains link speed samples, and finally synthesizes the speed samples into link speed and expresses it.

The bus link speed calculation method based on real-time bus GPS data provided by the present invention establishes a GPS sampling point queue for each vehicle in the speed sample acquisition stage, stores the matched GPS sampling points, and uses the time threshold to control the storage window size of the queue , and then establish a geofence with the station as the center, determine the key points used to calculate the speed sample according to the sorted sampling point information in the queue and the coordinate range of the geofence, and then extract the key point sequence of each bus to calculate the speed Sample; in the link speed synthesis stage, maintain its speed queue for each link, add the speed samples calculated in the previous stage to the speed queue of the link covered by the vehicle, and dynamically synthesize the link speed according to the number of queue samples , and finally average the velocity of multiple short-circuit links adjacent to each other as the final velocity of each link.

The present invention specifically includes the steps of speed sample acquisition and road link speed synthesis; the speed sample acquisition step includes maintaining the GPS sampling point queue, generating a sequence of key sampling points, and then obtaining speed samples; the road link speed synthesis step includes maintaining the road link The speed sample queue determines the number of speed samples used for calculation, and then synthesizes to obtain the link speed.

The maintenance GPS sampling point queue includes:

Create a queue of GPS sampling points for each bus, and each time a newest GPS point is inserted, it is judged whether the timestamp of the GPS point is the latest. ; Then judge whether the difference between the GPS point timestamps at the head of the queue and the tail of the queue meets the requirement of being less than the time threshold, and judge in a loop until the condition is met; finally update the sampling point information at the head of the queue.

Described generation key sampling point sequence comprises:

The thread for calculating the link speed is triggered every time period t _0. First, the GPS coordinates of each site are centered to delineate a geofence; the GPS sampling points in the queue are traversed, and the first one outside the range of the site geofence is selected. Sampling point, as the key sampling point used to calculate the speed of the bus; finally judge the sampling point of the head and tail of the queue, if it is outside the geographic fence range of the site, the sampling point is considered to be a key sampling point, and then each t ₀ The critical sampling points of a vehicle within a period generate a sequence of critical sampling points.

The velocity samples include:

Traverse the sequence of key sampling points, calculate every two sampling points as a pair, calculate the distance and time difference between the sampling points from the information of the key sampling points, and then apply the speed formula to get the speed between every two key sampling points, as the speed The sample is added to the speed sample queue of this link.

The speed sample queue step of maintaining the road link includes:

Maintain a speed sample queue for each link, update the queue information every time a new speed sample is inserted, and cyclically judge the timestamp information of each speed sample in the queue, and the sample data that exceeds the set time threshold from the current time will be removed from the queue removed from the .

The speed sample fusion based on k-nearest neighbor weighting includes:

For non-highway sections, compare the number of speed samples in the speed sample sets of time periods [0, T ₁ ], [T ₁ , T ₂ ], [T ₂ , T ^* ] with the size of the set thresholds N ₁ and N ₂ , where T ₁ and T ₂ are two time parameters for dividing the speed sample set, and T ₁ <T ₂ , T ^* is the time span between the taken speed sample set and the current moment (refer to Figure 3); determine by comparing the relationship Using one of the two velocity sample fusion methods based on k-nearest neighbor weighting, the set of velocity samples used for calculation is expressed as:

{V ₁ ,...,V _i ,...,V _k }

The first weighting scheme is the scheme that the closer the speed sample is to the real time, the greater the weight. t _i represents the absolute value of the difference between the i-th sample time and the current time. The weight formula is as follows:

The speed sample fusion formula is as follows:

In addition, different time proximity is divided into two different speed sample sets. The average method of equal weight is used in the set. The closer the time is, the larger the weight is followed between the sets, that is, the second weighting scheme. The number of speed samples is k ₁ and k ₂ respectively, and the weight formula is as follows:

The speed sample fusion formula is as follows:

For the high-speed section, the first weighting scheme is adopted.

The link speed synthesis includes:

For multiple connected road links of the same nature, if the connection between each road link is not a traffic light intersection, toll station, fork, etc., which have the effect of delaying vehicle travel, and the angle change α of the road link direction satisfies |α|≤α ₀ , α ₀ ＞0, where α ₀ is the angle threshold (the angle threshold is 30 degrees), then these connected links are regarded as a synthetic link, if this synthetic link is composed of n links, and Among them, m links have link speed results, respectively represented by {V ₁ ,...,V _i ,...,V _m }, and V _junc represents the speed of the synthesized link, then the link speed of the synthesized link for:

Compared with the prior art, the present invention has the advantages that: the method for calculating the speed of the bus link provided by the present invention, by generating a sequence of key sampling points and then calculating the speed samples, eliminates the influence of the speed of the bus stop on the road link within the range of the station. The calculated impact improves the accuracy of bus link speeds. The invention dynamically determines the sample size used for calculation according to the state of the speed sample queue, and performs dynamic calculation according to the real-time changing traffic conditions, and has better real-time performance and accuracy.

Description of drawings

Fig. 1 is the flow chart of the bus operation link speed calculation method based on real-time bus GPS data provided by the embodiment of the present invention;

Fig. 2 is a schematic diagram of the principle of generating a sequence of key sampling points in an embodiment of the present invention;

FIG. 3 is a schematic diagram of the principle of managing velocity sample sets according to time gradients in an embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.

The bus link speed calculation method based on real-time bus GPS data provided by the present invention establishes a GPS sampling point queue for each vehicle in the speed sample acquisition stage, stores the matched GPS sampling points, and uses the time threshold to control the storage window size of the queue , and then establish a geofence with the station as the center, determine the key points used to calculate the speed sample according to the sorted sampling point information in the queue and the coordinate range of the geofence, and then extract the key point sequence of each bus to calculate the speed Sample; maintain its speed queue for each road link in the link speed synthesis stage, add the speed samples calculated in the previous stage to the speed queue of the road link covered by the vehicle, and dynamically determine the number of queue samples used for calculation The number of speed samples, and determine the appropriate speed fusion scheme according to the distribution of the number of speed samples in different time ranges, and then obtain the synthetic speed of the link, and finally average the speed of multiple short-circuit links adjacent to the synthetic expression, Assigned to each link as its final velocity.

Fig. 1 is the flow chart of the bus link speed calculation method based on real-time bus GPS data provided by the embodiment of the present invention, comprising the following steps:

Step 101, speed sample acquisition.

Step 1011, maintaining the queue of GPS sampling points.

Establish a GPS sampling point queue for each bus, and store the matched and filtered GPS sampling points of the bus within T time. Each time a newest GPS point is inserted, it is judged whether its timestamp is the latest. If it is not the latest, no operation will be performed. If it is the latest, it will trigger to update the queue information, including the number of sampling points in the queue, the information of the sampling points at the end of the queue, and the queue. Update time, etc.; then judge whether the GPS point timestamp difference between the head of the queue and the tail of the queue meets the requirement of being less than the time threshold T. , T takes 15 minutes), then the GPS sampling point at the head of the queue is eliminated, and the loop judgment is performed until the condition is met, and finally the information of the sampling point at the head of the queue is updated.

Step 1012, generating a sequence of key sampling points.

The thread for calculating the link speed is triggered every time t ₀ passes. In order to eliminate the influence of stops on the overall driving speed of the bus, the present invention divides the queue of GPS sampling points, and then extracts the sequence of key points for calculating the driving speed of the bus between two points.

Figure 2 is a schematic diagram of extracting key sampling point sequences. First, take the GPS coordinates of each site as the center, and delineate a geographic fence with a radius of R. Traverse the GPS sampling points stored in chronological order in the queue, and select the first sampling point outside the geographic fence range of the site (considered to be the first sampling point before entering the station and after exiting the station), as the first sampling point used to calculate the bus travel The key sampling point of the speed; judge the sampling points of the head and the tail of the team, if the sampling point of the head or the tail of the team is within the geographical fence range of the site, it will not be regarded as a key sampling point, if it is outside the geographic fence range of the site, it will be Consider it a key sampling point.

For the step of generating the sequence of key sampling points from the sampling point queue of bus B within the T time range, the following formula can be obtained:

{Δs ₁ ,Δs ₂ ,...,Δs _n |T,B}=sp(Δs|T,B);

{Δt ₁ ,Δt ₂ ,...,Δt _n |T,B}=sp(Δt|T,B);

Among them, Δs and Δt are the distance and time difference calculated from the information carried by the sampling points. The function of the sp function is to cut the queue of sampling points with the station, and then obtain the distance and time difference information between a series of key sampling points. The generated sequence of key sampling points is used to calculate the speed samples of each vehicle within the driving range and applied to each corresponding road link. Step 1013, calculating velocity samples. Use pairs of keypoints to calculate the speed of a transit vehicle between two points.

Traverse the sequence of key sampling points, and calculate every two sampling points as a pair. The distance between sampling points is calculated from the GPS information of key sampling points, the time difference between sampling points is calculated from the time stamp information of key sampling points, and the speed v _i between every two key sampling points is obtained by applying the speed formula.

The calculated driving speed between key sampling points is added to the speed sample queue of all links passing between two sampling points. Each velocity sample is used by multiple links to predict link velocity.

Step 102, link speed synthesis.

Step 1021, maintain the speed sample queue of the link.

Maintain its speed sample queue for each road link, add the speed samples calculated in the previous stage to the speed queue of the road link covered by the vehicle, and update the queue information every time a new speed sample is inserted. The real-time GPS of the bus vehicles has a time difference, so the speed samples in the queue are not sorted. The time window size of the link speed sample queue is T ^* . In each calculation cycle with an interval of t ₀ , the time stamp information of each speed sample in the queue is cyclically judged. The sample data that is more than T ^* from the current time will be removed from the queue. remove.

Step 1022, determine the sample size used for calculation.

For road sections of different nature, the schemes to obtain the sample size for calculation are different. For non-highway sections, the sample size is selected according to the following rules.

Fig. 3 is a schematic diagram of the principle of managing the speed sample set according to the time gradient.

Speed samples within T ^* time are stored in the speed sample queue, and the speed sample queue is divided into three gradients according to the time threshold, which are [0,T ₁ ], [T ₁ ,T ₂ ], [T ₂ , T ^* ] velocity sample set, where T ₁ and T ₂ are two time thresholds for dividing the velocity sample set, and T ₁ < T ₂ , T ^* is the time span between the selected velocity sample set and the current moment, and its The number of included samples is n ₁ , n ₂ , n ₃ respectively, and the two thresholds for setting the number of samples are N ₁ and N ₂ , where N ₁ < N ₂ , it is necessary to determine the number of speed samples k ₁ and k for calculation ₂ .

First judge the number n ₁ of speed samples in [0, T ₁ ], if n ₁ ≥ N ₁ , it means that the road section is densely packed with vehicles, select n ₁ speed samples in 0-T ₁ for calculation, and select the second weighting Scheme, let k ₁ =n ₁ , continue to step 1023;

If n ₁ <N ₁ , it means that there are few speed samples in a short period of time, which is not representative. Consider expanding the speed sample set; judge the number of speed samples in [0,T ₂ ] n ₁ +n ₂ , if n ₁ +n ₂ >N ₂ , select n ₁ +n ₂ speed samples in [0, T ₂ ] for calculation, select the second weighting scheme, set k ₁ =n ₁ , k ₂ =n ₂ , and continue to step 1023;

If N ₁ <n ₁ +n ₂ <N ₂ , select the first weighting scheme, set k ₁ =n ₁ +n ₂ , and continue to step 1023;

If n ₁ +n ₂ <N ₁ , and n ₁ +n ₂ +n ₃ >N ₁ , select the first weighting scheme, set k ₁ =n ₁ +n ₂ +n ₃ , and continue to step 1023;

If n ₁ +n ₂ +n ₃ <N ₁ , the link is not calculated.

For the high _- speed road section, it has the characteristics of few parking stops, less dense buses, and relatively stable bus speeds. Therefore, the embodiment of the present invention selects n speed samples within T2 for calculation, and selects the first weighting scheme, so that k ₁ =n, go to step 1023 .

Step 1023, speed sample fusion based on k-nearest neighbor weighting.

The k samples obtained in step 1022 are already spatially adjacent to the speed of the predicted link, so the distribution of weights is considered in terms of temporal proximity, and different k-proximity weighting schemes are adopted due to the different sample sizes selected. The computed set of velocity samples is expressed as:

{V ₁ ,...,V _i ,...,V _k }

In the first weighting scheme, the closer the speed sample is to the real time, the greater the weight, and the weight is inversely proportional to the time difference between the sample time and the current time. t _i represents the absolute difference between the i-th sample time and the current time. value, the weight formula is as follows:

The speed sample fusion formula is as follows:

In the second weighting scheme, the speed samples are divided into different speed sample sets according to different time proximity, and the average method of equal weight is used in the set, and the principle that the closer the time is, the greater the weight is followed between the sets. The set of velocity samples used for calculation is expressed as:

{V ₁ ,...,V _i ,...,V _k1 }

{V _k1+1 ,...,V _j ,...,V _k2 }

The scheme weight formula is as follows:

The speed sample fusion formula is as follows:

Step 1024, combining speeds of adjacent links.

In order to meet the real-time calculation requirements of the bus link speed and the integrity of the results, the present invention proposes a method for synthesizing the speed of adjacent links to solve this problem. For multiple connected road links of the same nature, if the connection between each road link is not a traffic light intersection, a toll station, and a fork, which have the effect of delaying vehicle travel, and the angle change α of the road link direction satisfies |α|≤α ₀ (α ₀ ＞0), where α ₀ is the angle threshold, then these connected links are regarded as a synthetic link, if the synthetic link is composed of n links, and m links have path The link speed results are respectively expressed by {V ₁ ,...,V _i ,...,V _m }, and V _junc represents the speed of the synthesized link, then the link speed of the synthesized link is:

In a word, the bus link speed calculation method provided by the present invention eliminates the influence of bus stops on the link speed calculation within the station range by generating a sequence of key sampling points and then calculating the speed samples, and improves the bus link speed. the accuracy. The invention dynamically determines the sample size used for calculation according to the state of the speed sample queue, and performs dynamic calculation according to the real-time changing traffic conditions, and has better real-time performance and accuracy.

The above embodiments are provided only for the purpose of describing the present invention, not to limit the scope of the present invention. The scope of the invention is defined by the appended claims. Various equivalent replacements and modifications made without departing from the spirit and principle of the present invention shall fall within the scope of the present invention.

Claims (6)

1. A method for calculating the speed of public transport links based on real-time bus GPS data, characterized in that: comprise speed sample acquisition and road link speed synthesis steps; described speed sample acquisition steps include maintaining the GPS sampling point formation, generating key sampling points sequence, and then obtain speed samples; the link speed synthesis step includes maintaining the speed sample queue of the link, determining the speed sample size for calculation, and then synthesizing to obtain the link speed; The maintenance GPS sampling point queue includes: Create a queue of GPS sampling points for each bus, and each time a newest GPS point is inserted, it is judged whether the timestamp of the GPS point is the latest. ; Then judge whether the GPS point timestamp difference between the head of the team and the tail of the team meets the requirement of being less than the time threshold, and judge in a loop until the condition is met; finally update the sampling point information of the head of the team; Described generation key sampling point sequence comprises: The cycle of calculating the link speed is t ₀ , firstly, the GPS coordinates of each site are used as the center to delineate the geographic fence; the GPS sampling points in the queue are traversed, and the first sampling point outside the geographic fence range of the site is selected as the The key sampling point is used to calculate the speed of the bus; finally judge the sampling point of the head and the tail of the queue, if it is outside the geographic _fence range of the station, the sampling point is considered to be a key sampling point, and then the vehicle’s Key Sample Points Generates a sequence of key sample points. 2. the bus link speed calculation method based on real-time bus GPS data according to claim 1, is characterized in that: described speed sample comprises: Traverse the sequence of key sampling points, calculate every two sampling points as a pair, calculate the distance and time difference between the sampling points from the information of the key sampling points, and then apply the speed formula to get the speed between every two key sampling points, as the speed The sample is added to the speed sample queue of this link. 3. the bus link speed calculation method based on real-time bus GPS data according to claim 1, is characterized in that: the speed sample formation step of described maintenance link comprises: Maintain a speed sample queue for each link, update the queue information every time a new speed sample is inserted, and cyclically judge the timestamp information of each speed sample in the queue, and the sample data that exceeds the set time threshold from the current time will be removed from the queue removed from the . 4. the bus link speed calculation method based on real-time bus GPS data according to claim 1, is characterized in that: described link speed synthesis adopts the speed sample fusion based on k nearest neighbor weighting to realize, and described based on k nearest neighbor weighting The speed sample fusion includes: For non-highway sections, compare the number of speed samples in the speed sample sets of time periods [0, T ₁ ], [T ₁ , T ₂ ], [T ₂ , T ^* ] with the size of the set thresholds N ₁ and N ₂ , where T ₁ and T ₂ are two time parameters for dividing the velocity sample set, and T ₁ <T ₂ , T* is the time span between the velocity sample set and the current moment, and V _i represents the i-th velocity sample; the first A weighting scheme is a scheme in which the closer the speed sample is to the real time, the greater the weight. t _i represents the absolute value of the difference between the i-th sample time and the current time, the number of samples is k ₁ , and the weight formula is as follows : The speed sample fusion formula is as follows: In addition, different time proximity is divided into two different speed sample sets, and the average weighting method of equal weight is used in the set, the number of speed samples in the set is k ₁ , k ₂ respectively, and the time closer weight is followed between sets. The scheme with a larger value, that is, the second weighting scheme, the weight formula is as follows: The speed sample fusion formula is as follows: For the high-speed section, the first weighting scheme is adopted. 5. the bus link speed calculation method based on real-time bus GPS data according to claim 1, is characterized in that: described link speed synthesis comprises: for a plurality of connected link of same nature, if connect each road The chain is not a traffic light intersection, a toll booth, or a fork, which has the effect of delaying vehicle travel, and the angle change α of the road chain direction satisfies |α|≤α ₀ , α ₀ >0, where α ₀ is the angle threshold, then Consider these connected links as a composite link. If the composite link is composed of n links, and m links have link speed results, use {V ₁ ,...,V _i ,...,V _m } represent, V _junc represents the speed of the synthetic link, then the link speed of the synthetic link is: . 6. the bus link speed calculation method based on real-time bus GPS data according to claim 5, characterized in that: the angle threshold is 30 degrees.