WO2020244288A1

WO2020244288A1 - Method and apparatus for evaluating truck driving behaviour based on gps trajectory data

Info

Publication number: WO2020244288A1
Application number: PCT/CN2020/081355
Authority: WO
Inventors: 李颖; 杨临涧; 安毅生; 慕晨; 徐悦
Original assignee: 长安大学
Priority date: 2019-06-03
Filing date: 2020-03-26
Publication date: 2020-12-10
Also published as: CN110197588B; CN110197588A

Abstract

A method and an apparatus for evaluating truck driving behaviour based on GPS trajectory data, the method comprising the following steps: filtering all of the driving trajectory data of a vehicle to be measured from raw GPS data (101); processing the driving trajectory data to acquire valid driving trajectory data (102); analysing the valid driving trajectory data to obtain a valid study trajectory (103); and calculating a risky driving behaviour index for the valid study trajectory to evaluate the driving behaviour (104). The present method mines a large amount of GPS raw data to extract valid trajectory travel and identify the main routes of vehicle operation, and can accurately restore driving trajectory data by means of establishing a driving behaviour trajectory scoring mechanism for truck drivers with different indices and different weighting, increasing the precision of splitting travel trajectories, correctly proposing driving safety evaluation parameters, and evaluating potential risky driving behaviour and excessive driving loads.

Description

Method and device for evaluating large truck driving behavior based on GPS trajectory data

Technical field

The invention relates to the technical field of road safety, in particular to a method and device for evaluating the driving behavior of a large truck based on GPS trajectory data.

Background technique

In recent years, the accident rate of large trucks in my country has increased year by year. This is because large trucks have frequent traffic accidents due to poor braking performance, overloading and overspeeding, and fatigue driving. Therefore, the impact of trucks on traffic safety has become an urgent problem to be solved.

With the large-scale installation of GPS satellite positioning devices on large trucks, the use of track data recorded during driving makes it possible to study the driving behavior patterns of large trucks. During operation, the GPS device will record spatial position (latitude and longitude), time, speed and other data at a constant sampling rate. Compared with the traditional small-scale single-point data collection and analysis, massive GPS trajectory data has many advantages in studying the safety of driving behavior of large trucks during operation. However, without additional information, there are still huge challenges in analyzing and processing large-scale trajectory data in spatiotemporal environments. In related technologies, the difficulties of GPS trajectory data processing mainly focus on the cleaning of system errors and random noise, the segmentation of a single trip, and the extraction of effective traffic parameters in the trajectory sequence.

In the above technologies, only systematic errors and random noises are cleaned up, and the processing of the obtained GPS trajectory data cannot finally obtain an accurate evaluation plan, which results in the evaluation method deviating from the actual situation, and driving behavior research cannot be accurately conducted. Therefore, how to use these massive GPS data to find a set of suitable post-processing methods to effectively extract key information is very important for the accuracy of subsequent driving safety research.

Summary of the invention

In view of the shortcomings of the above-mentioned prior art, the present invention provides a method and device for evaluating the driving behavior of a large truck based on GPS trajectory data, which can accurately restore driving trajectory data, improve the accuracy of dividing itinerary trajectories, and correctly propose driving safety evaluation parameters. Provide a basis for driving safety research based on massive trajectory data.

The technical scheme adopted by the present invention is:

A method for evaluating the driving behavior of a large truck based on GPS trajectory data, including the following steps:

Filter all the driving track data of the vehicle to be tested from the original GPS data;

Process the driving track data to obtain effective driving track data;

Analyze and process the effective driving trajectory data to obtain an effective research trajectory;

Calculate the risky driving behavior index of the effective research trajectory to evaluate the driving behavior.

As a further technical solution of the present invention, the selection of all driving trajectory data of the vehicle to be tested from the original GPS data is specifically: extracting the vehicle to be tested and the number of the GPS device installed on the vehicle to be tested Test all the driving track data of the vehicle.

As a further technical solution of the present invention, the processing of the driving trajectory data to obtain effective driving trajectory data includes:

Preprocessing the driving track data to obtain the first processed data;

Perform segmentation processing on the first processed data to determine effective travel trajectory data.

As a further technical solution of the present invention, the preprocessing of the driving trajectory data to obtain the first processed data specifically includes: clearing abnormal data and duplicate data in the driving trajectory data.

As a further technical solution of the present invention, the segmentation processing of the first processed data to determine the effective travel trajectory data specifically includes:

Filter the first processed data by band-pass filtering algorithm and filter the speed duration and speed interval time;

Compare the speed duration and speed interval with the preset minimum effective stroke length threshold;

The trajectory data corresponding to the speed duration and the speed interval that are greater than the minimum length threshold of the effective stroke are extracted to obtain the effective stroke trajectory data.

As a further technical solution of the present invention, the first processed data is filtered through a band-pass filtering algorithm and the speed duration and speed interval are selected; specifically including:

Determine the error value of the first processed data through a band-pass filtering algorithm, and remove the error value to form the second processed data;

Filter the second processed data to determine the speed duration and the speed interval, and determine whether there is a data gap;

If there is a data gap, judge the data gap time, if the data gap time reaches the set threshold, mark it as a candidate track;

Otherwise, the average interpolation method is used for data restoration.

As a further technical solution of the present invention, the second processing data is screened to determine the speed duration and the speed interval time, and to determine whether there is a data gap; specifically:

By comparing the recorded interval time and calculating the distance between the data points before and after the interval and the average speed quotient, it is determined whether there is a data interval caused by signal loss in the recording interval.

As a further technical solution of the present invention, the effective driving track data is analyzed and processed to obtain an effective research track; specifically including:

Perform similarity measurement and cluster analysis on effective travel trajectory data;

Match all valid travel trajectories according to their latitude and longitude coordinates;

Determine the route with more repetitions as an effective research trajectory.

As a further technical solution of the present invention, the similarity measurement and cluster analysis on the effective travel trajectory data specifically include:

Choose two travel trajectories from the effective travel trajectory data, and randomly extract some data points for the two trajectories to calculate the similarity of the two travel trajectories;

Calculate the similarity between all trajectories in the effective travel trajectory data, and finally obtain the similarity matrix of all trajectories;

Calculate the variance of clusters between different trajectories to determine the optimal number of clusters.

As a further technical solution of the present invention, the calculation of the risky driving behavior index of the effective research trajectory to evaluate the driving behavior specifically includes:

According to the driving trajectory data corresponding to the optimal number of clusters and the increase in driving load, the driving behavior is evaluated.

As a further technical solution of the present invention, the driving behavior is evaluated according to the driving trajectory data corresponding to the optimal number of clusters and the increase in driving load, specifically:

Determine risk driving indicators; including the frequency of speeding and the frequency of sudden acceleration and deceleration;

Determine the driving load index; this index uses the break rate to calculate the break rate of the vehicle to be tested;

Combining the above two indicators, the risk driving of a single driver is scored; the calculation formula is as follows:

Among them, Fp _i is the comprehensive score of the driving behavior trajectory of a large truck driver i; β(n) is the weight coefficient of different indicators; F ^v is the frequency at which the instantaneous speed of the vehicle under test exceeds the preset safety threshold, and F ^a is the test vehicle The frequency at which the acceleration of the vehicle exceeds the preset safety threshold; SD/SG is the break rate, which is the ratio of the total driving time to the total stop time.

The present invention also provides a large truck driving behavior evaluation device based on GPS trajectory data, including:

The data screening unit is used to screen all the driving track data of the vehicle to be tested from the original GPS data;

The first processing unit is configured to process the driving trajectory data to obtain effective driving trajectory data;

The second processing unit is used to analyze and process the effective driving trajectory data to obtain an effective research trajectory;

The driving behavior evaluation unit is used to calculate the risk driving behavior index of the effective research trajectory to evaluate the driving behavior.

The beneficial effects of the present invention are:

The invention excavates massive GPS raw data, extracts the effective trajectory of the truck, and recognizes the main route of the vehicle; through the statistical analysis of the vehicle's multiple round trips on the same route, different measurement indicators are used to characterize the potential Establishing a scoring mechanism for driving behavior trajectory of large truck drivers with different indicators and different weights, which can accurately restore driving trajectory data, improve the accuracy of dividing itinerary trajectories, and correctly propose driving safety assessment Parameters to evaluate potential risky driving behavior and excessive driving load.

Description of the drawings

Figure 1 is a flow chart of a method for evaluating the driving behavior of a large truck based on GPS trajectory data proposed by the present invention;

Figure 2 is a flow chart for acquiring effective driving track data proposed by the present invention;

Figure 3 is a flow chart for obtaining effective research trajectories proposed by the present invention;

Figure 4 is a flowchart of a specific embodiment proposed by the invention;

Figure 5 is a structural diagram of a large truck driving behavior evaluation device based on GPS trajectory data proposed by the present invention;

Figure 6a is a speed trajectory diagram before filtering according to an embodiment of the present invention;

Fig. 6b is a speed trajectory diagram after LMS filtering in the present invention;

Figure 7a is an error rate indication diagram in a driving state according to an embodiment of the present invention;

Fig. 7b is an error rate indication diagram in a stopped state according to an embodiment of the present invention;

Figure 8a is a diagram of all travel trajectories of the vehicle to be tested;

Figure 8b is a travel trajectory diagram on the research route of the vehicle to be tested;

Figure 9a is an actual trajectory and speed change pattern diagram of an embodiment of the present invention;

Figure 9b is a diagram of the actual trajectory and speed change pattern of an embodiment of the present invention;

Fig. 9c is a diagram of the actual trajectory and speed change pattern of an embodiment of the present invention;

Fig. 9d is a diagram of the actual trajectory and speed change pattern of an embodiment of the present invention.

Detailed ways

The invention provides a method for evaluating the driving behavior of a large truck based on GPS trajectory data. It extracts the trajectory data of the vehicle to be tested from the massive GPS trajectory data, obtains the main route of the vehicle through filtering and calculation, and extracts the driving according to the main route of the vehicle. Parameters used to evaluate driving behavior.

The general idea of the technical solution provided by the present invention is as follows:

The invention provides a method for evaluating the driving behavior of a large truck based on GPS trajectory data; it is used for vehicle supervision and driver behavior analysis; massive GPS raw data is excavated, effective trajectory travel is extracted, and the main route of vehicle operation is identified. Through the statistical analysis of multiple round trips of vehicles on the same route, different measurement indicators are used to characterize potential driving behavior risks, and quantitative risk coefficients are given for the evaluation of driving behavior.

The above is the core idea of the application. In order to enable those skilled in the art to better understand the solution of the application, the application will be further described in detail below in conjunction with the accompanying drawings. It should be understood that the embodiments of the application and the specific features in the embodiments are detailed descriptions of the technical solutions of the application, rather than limitations on the technical solutions of the application. In the case of no conflict, the embodiments of the application and the technical features in the embodiments Can be combined with each other.

Example one

As shown in Fig. 1, it is a flowchart of a method for evaluating the driving behavior of a large truck based on GPS trajectory data proposed by the present invention.

Referring to Figure 1, a method for evaluating the driving behavior of a large truck based on GPS trajectory data includes the following steps:

Step 101: Filter all driving track data of the vehicle to be tested from the original GPS data;

Step 102, processing the driving trajectory data to obtain effective driving trajectory data;

Step 103: Analyze and process the effective driving trajectory data to obtain an effective research trajectory;

Step 104: Calculate the risky driving behavior index of the effective research trajectory to evaluate the driving behavior.

The invention excavates massive GPS raw data, extracts effective trajectory distance, and identifies the main route of vehicle operation. Through the statistical analysis of multiple round trips of vehicles on the same route, different measurement indicators are used to characterize potential driving behavior risks, and quantitative risk coefficients are given; the driving behavior trajectories of large truck drivers with different indicators and different weights are established The scoring mechanism evaluates potential risky driving behavior and excessive driving load.

In step 101, all the driving trajectory data of the vehicle to be tested are deleted from the original GPS data, specifically, all the driving trajectory data of the vehicle to be tested is extracted according to the number of the vehicle to be tested.

In the embodiment of the present invention, all trajectory data of the vehicle to be tested are extracted from the original GPS data of the database or the data platform, and are labeled and numbered. Since the vehicle has a unique vehicle number (ie license plate), all data of the vehicle can be accurately extracted through the vehicle number, and all the GPS data obtained during the driving of the vehicle to be tested are extracted as driving track data. It also decodes and organizes the driving trajectory data, such as encoding the system time, unifying the format, and determining the data point interval.

Refer to FIG. 2, which is a flowchart of obtaining effective driving track data proposed by the present invention;

As shown in Figure 2, in step 102, processing the driving trajectory data to obtain effective driving trajectory data specifically includes:

Step 121: Preprocess the driving trajectory data to obtain first processed data;

Step 122: Perform segmentation processing on the first processed data to determine effective travel trajectory data.

Wherein, preprocessing the driving trajectory data to obtain the first processed data specifically includes: clearing abnormal data and duplicate data in the driving trajectory data.

In the embodiment of the present invention, the driving track data is cleaned, and abnormal data and duplicate data are first eliminated through data preprocessing. Abnormal data refers to a value that is actually impossible to achieve. For example, if the speed of a large truck exceeds 200km/h, it is regarded as abnormal data; a negative speed value can also be regarded as abnormal data; duplicate data refers to multiple identical records at the same time point The data. Both of these situations are caused by system errors. The preprocessing stage is processed first, which helps the accuracy and speed of the next filtering stage.

The present invention regards GPS trajectory data as a signal wave. In a movable time window with a width of 2m and a center at time t, the signal output by the GPS device can be represented by the following model:

fori=-m,...,m;

among them,

Represents the basic signal, that is, the true signal component, and its change is generally relatively stable without sudden changes; β _t represents the slope of the center of the time window, ε _{t, i} represents the noise component at the i-th moment, which may contain due to Outliers caused by measurement errors or random errors; the residual in the time window t can be expressed as:

for i=-m,...,m;

The key to the filtering algorithm is to estimate

And β _t value, the adaptive least median square (Least Median of Squares, LMS) filtering algorithm used in the present invention is estimated as follows:

for i=-m,...,m;

The present invention uses the LMS algorithm to filter the driving trajectory data. According to the characteristic that the speed trajectory data of vehicles (especially large trucks) will not rise or fall suddenly, the extracted driving trajectory data can be regarded as signal data with noise. The speed trajectory is smoothed by an adaptive Least Median of Squares (LMS) filtering algorithm. The LMS filtering algorithm is a robust regression method that uses a moving median value, is more robust to outliers, can follow changes in the trajectory, and is not affected by abnormal signals. Through filtering and denoising, the misjudgment of effective travel data is greatly reduced.

In step 122, performing segmentation processing on the first processed data to determine effective travel trajectory data includes:

Extract the trajectory data corresponding to the speed duration and the speed interval that are greater than the minimum length threshold of the effective stroke to obtain the effective stroke trajectory data.

The continuous GPS data is segmented by the band-pass filtering algorithm and two indicators (speed duration and speed gap), and the minimum effective travel length threshold is determined according to actual research needs. For large trucks running on the main road, an effective running time, that is, the speed lasts no less than 15 minutes; similarly, the present invention also considers an effective stop time, that is, the speed interval is no less than 15 minutes. In addition, the speed duration in the present invention measures the time period when the continuous GPS speed data points are greater than 5km/h, and the speed interval time measures the time period when the continuous GPS speed data points are less than 5km/h (that is, the stopping speed). . When the two time periods are respectively greater than the preset time threshold (for example, 15 minutes), an independent trip can be divided. The travel trajectory required by the research is screened by the above method.

In the above steps, the first processed data is filtered through a band-pass filtering algorithm and the speed duration and speed interval are filtered; specifically including:

Otherwise, the average interpolation method is used for data restoration.

Assuming the fluctuating trajectory curve obtained by the vehicle in a stable running state, the smaller fluctuations are generally within the accuracy allowable range of the measuring instrument. If the fluctuation value exceeds the vehicle performance range, it may be caused by measurement errors. For example, in the normal operation of a large truck, the GPS data shows sudden changes (sudden rise or drop) in the speed of two adjacent moments or within two seconds before and after. Therefore, the errors that will be generated by the filtering algorithm are also divided into the following two categories:

Category 1: speed dwell (SD) means the speed is continuous, that is, the vehicle is driving. In this state, the vehicle speed should remain within the current speed limit and its fluctuation range. Set the driving speed S (km/h). Data points whose instantaneous speed drops to 5km/h and below will be smoothed out by the filtering algorithm. Count the data points n smoothed out by the filtering algorithm in this process. The smoothing error of each trip i in the driving state is calculated as

Among them, N is the total number of data points of the trip i.

Category 2: speed gap (SG) is the speed gap, that is, the vehicle is at a standstill state. In this state, the vehicle speed is less than or equal to 5km/h; the data points whose instantaneous speed rises to S and above will be smoothed by the filtering algorithm; statistics of the process The data points m smoothed out by the filtering algorithm in, the smoothing error of each stroke j in the resting state is calculated as εSG _{, j} = m _j /M _j ; where M is the total number of data points in the interval j.

Further, the second processing data is screened to determine the speed duration and the speed interval, and determine whether there is a data discontinuity; specifically:

The present invention determines whether there is a data discontinuity caused by signal loss in the recording discontinuity by comparing the recorded discontinuous time interval and calculating the quotient of the distance between the data points before and after the interval and the average speed, that is, whether a certain travel trajectory is complete; and through the average interpolation method Perform small-range data repair; if the interruption time is too long (for example, more than one-tenth of the entire trip, it is marked as an alternative trajectory for the trip). In this way, the data integrity of a trip is classified, and the trajectory with higher integrity is preferred for analysis.

Participate in Figure 3, which is a flow chart for obtaining effective research trajectories proposed by the present invention;

As shown in Figure 3, in step 103, the effective driving track data is analyzed and processed to obtain an effective research track; specifically including:

Step 131: Perform similarity measurement and cluster analysis on the effective travel trajectory data;

Step 132: Match all valid travel trajectories according to their latitude and longitude coordinates;

Step 133: Determine a route with more repetitions.

Since the operation and driving behavior of vehicles are greatly affected by different road conditions and traffic environment, the comparability between different routes is not strong; for large trucks, they travel between A and B in a short time, and the transportation route Relatively single; through trajectory similarity measurement and cluster analysis, all valid travel trajectories are matched according to their latitude and longitude coordinates, and the routes with more repetitions can be determined. Therefore, the present invention extracts the route with the highest repetition rate of vehicle operation to study and compare the driving behavior of the vehicle in different operating states on the route.

Refer to Fig. 4 for a flowchart of a specific embodiment proposed by the invention; specifically, the research route is realized through trajectory similarity and cluster analysis.

As shown in Figure 4, the similarity measurement and cluster analysis of the effective travel trajectory data include:

Choose two travel trajectories from the effective travel trajectory data, and measure the similarity of the two trajectories;

Calculate the change of cluster variance between different trajectories to determine the optimal number of clusters.

Due to the fuzzy classification boundary of the data set itself, the number of clusters is subject to a certain degree of subjective judgment standards; a common method uses hierarchical clustering to check the output of the dendrogram to determine the optimal number of clusters. The present invention uses the change of cluster variance between different trajectories to determine the optimal number of clusters; the trajectory cluster with the highest similarity is determined as the research route; all trajectories in the cluster are used as the basic data source for evaluating the driving behavior of large truck drivers ; At the same time remove all tracks not on the research route.

Specifically, two travel trajectories are selected from the divided travel data set, and some data points of the two trajectories are randomly selected for calculation to reduce the amount of calculation.

Similarity measure: Calculate the longest common subsequence. The present invention uses the most effective method based on the longest common subsequence (LCSS) to calculate the degree of similarity between trajectories; the value of LCSS is the length of the longest matching subsequence that can be obtained by matching the data points on the two trajectories P and Q, The algorithm can be expressed by the following formula:

The main idea of LCSS is to iteratively calculate the distance between the data point p and the data point q on the two trajectories, and compare the preset value ε to judge whether the two data points match. The algorithm allows two trajectories to stretch, compress, twist, and mismatch individual data points, making it easy to process some low-quality trajectory data. The main advantages of this method are: (1) allow trajectory extension, compression, warping, etc. (elasticity), (2) allow trajectories of different lengths (time shift), (3) insensitive to error data points (robustness) ).

Calculate the similarity of two trajectories: After the longest common subsequence is obtained by calculation, the similarity of the two trajectories can be expressed by the following formula:

S _LCSS (P, Q)=LCSS(P, Q)/min(n, m);

Where n and m are the data point lengths of the two trajectories P and Q; SLCSS (P, Q) is between 0 and 1; by definition, the closer the value of SLCSS (P, Q) is to 1, it means the two trajectories The more similar, the less similar the two trajectories.

Calculate the similarity between all trajectories in the data set, and finally obtain the similarity matrix of all trajectories.

Cluster analysis: calculate the optimal number of clusters. Because the classification boundary of the data set itself is relatively fuzzy, the number of clusters has a subjective judgment standard to a certain extent. A common method is to use hierarchical clustering to check the output of the dendrogram to determine the optimal number of clusters. In the present invention, the cluster variance value changes between different trajectories are used to determine the optimal number of clusters. When the trajectories are correctly classified into different corresponding clusters, the variance between similar trajectories should be small, and the variance between different types of trajectories should be large.

Use cluster variance as the index of classification. Cluster variance represents the coordinate square deviation of the cluster mean of all observations in the cluster, including within-cluster Sum-of-Squares (WSS) and between-cluster Sum-of-Squares , BSS). The calculation formulas are as follows:

The within-cluster variance WSS is used to measure the variance within each cluster, which is the sum of squared distances from all trips in the cluster to the centroid of the class trips. Generally, the intra-cluster sum of squares with a small sum of squares is tighter than the intra-cluster sum of squares with a large sum of squares, that is, the better the cluster, the smaller the overall WSS. The inter-cluster variance BSS is used to measure the variance between clusters. It is the sum of the squared distances between all the strokes of each cluster and the centroid stroke. The optimal number of clusters should be that when a new cluster is added, the value of the sum of squares within the cluster will not change much. The larger the BSS, the better the clustering result. The total variance is the sum of the sum of squares within a cluster and the sum of squares between clusters. The goal of the clustering algorithm is to minimize the squares within the cluster and maximize the sum of squares between the clusters. To optimize and determine the optimal number of clusters.

Carry out hierarchical clustering analysis according to the degree of similarity of trajectories, and the trajectories with higher similarity (>95%) constitute the candidate route; classify all trajectory data to obtain the route classification with the largest number of trajectories. All trajectories on this route will be used as data sources for the next analysis.

In step 104, the risk driving behavior index of the effective research trajectory is calculated to evaluate the driving behavior, which specifically includes:

Wherein, the evaluation of the driving behavior according to the driving trajectory data corresponding to the optimal cluster number and the increase of the driving load is specifically:

Determine the driving load index; this index uses the break rate to calculate the total driving time and total stop time of the vehicle to be tested per day (24 hours). The break rate is the ratio of the total driving time to the total stop time;

Combining the above two indicators, the risk driving of a single driver is scored.

Many indicators can be used to measure driver behavior assessment related to potential or actual risks of accidents, including speed, acceleration, jerk, lane change, driving load, driving duration, braking frequency, etc. GPS can provide complete trajectory data before, during, and after dangerous driving behaviors (such as sudden braking), so that the stability of driving behavior can be well evaluated. The driving behavior of large trucks is mainly potential risky driving behavior, which is reflected in the driving trajectory data, including sudden changes in speed and acceleration, as well as increased driving load.

The purpose of the present invention is to evaluate potential risky driving behavior and excessive driving load, especially a method of large-scale monitoring and regular evaluation of large truck drivers. Based on the limited data information, the traffic parameter index of the large truck under multiple round-trip operations on the same route is extracted, and the driving behavior trajectory scoring mechanism of the large truck driver with different indicators and different weights is established. The present invention mainly uses risk driving indicators and driving review indicators. Evaluation of driving behavior.

Determining risk driving indicators: The study found that drivers with less risk are associated with smooth operating modes, while unstable and aggressive operating modes, such as sudden increases and decreases in speed or frequent accelerations and decelerations, are significantly related to the driver’s accident rate . The present invention is extracted speeding frequency and the frequency of sudden acceleration or deceleration as a potential risk driving behavior index, calculated each vehicle's instantaneous velocity and acceleration exceeds a preset safety threshold frequency F ^v and F ^a.

Determine the driving load index: Calculate the total driving time and total stop time of each vehicle per day (24 hours), and calculate the driving break rate SD/SG. The greater the ratio, the greater the driving load.

Combining the above two indicators, a single driver is scored for risky driving; the higher the score, the higher the probability of risky driving behavior. Calculated as follows:

Among them, Fp _i is the comprehensive score of the driving behavior trajectory of a large truck driver i; β(n) is the weight coefficient of different indicators. Because different indicators have different degrees of importance or contribution to the assessment of driving risk, users can weight each indicator according to the importance or severity. The higher the driving risk, the higher the corresponding weight.

The invention is used for vehicle supervision and driver behavior analysis. Its beneficial effect is to mine massive GPS raw data, extract effective trajectory travel, and identify the main route of vehicle operation. Through the statistical analysis of multiple round trips of vehicles on the same route, different measurement indicators are used to characterize potential driving behavior risks, and quantitative risk coefficients are given.

Example two

Based on the same inventive concept as the method for evaluating the driving behavior of a large truck based on GPS trajectory data in the foregoing embodiments, the present invention also provides a device for evaluating driving behavior of a large truck based on GPS trajectory data.

Referring to Figure 5, a large truck driving behavior evaluation device based on GPS trajectory data includes:

The data screening unit 201 is used for screening all driving track data of the vehicle to be tested from the original GPS data;

The first processing unit 202 is configured to process the driving trajectory data to obtain effective driving trajectory data;

The second processing unit 203 is configured to analyze and process the effective driving trajectory data to obtain an effective research trajectory;

The driving behavior evaluation unit 204 is used for calculating the risk driving behavior index of the effective research trajectory to evaluate the driving behavior.

Various changes and specific examples of the method for evaluating the driving behavior of a large truck based on GPS trajectory data in the first embodiment are also applicable to the device for evaluating driving behavior of a large truck based on GPS trajectory data of this embodiment. A detailed description of a method for evaluating the driving behavior of a large truck based on GPS trajectory data. Those skilled in the art can clearly know a device for evaluating the driving behavior of a large truck based on GPS trajectory data in this embodiment. This will not be detailed here.

Example three

In this embodiment, GPS data sets of several large trucks are selected as the research objects. Based on the detailed steps of the above research content, briefly list the key steps for explanation.

1. The realization process of itinerary trajectory extraction:

Extract all trajectory data of any vehicle from the database or data platform, and label and number them.

The trajectory data is decoded and sorted, and the original data is preliminary sorted. For example, to encode the system time, unify the format, and determine the data point interval.

Based on the characteristics that the speed trajectory data of vehicles (especially large trucks) will not rise or fall sharply, the original speed trajectory data can be regarded as signal data with noise. Through filtering and denoising, the misjudgment of effective stroke is greatly reduced. The black curve in Fig. 6a is the speed trajectory before filtering, and the black curve in Fig. 6b is the speed trajectory after LMS filtering.

The effective travel trajectory can be regarded as a band-pass signal, and the dynamic bandwidth allows all speed trajectories with a speed greater than 5km/h to pass. The straight line in Figure 6a and Figure 6b is the band-pass filter, allowing valid signals to pass and blocking invalid signals (stop speed).

Obtain an effective trajectory and research route. Figure 7 shows the error rate in the driving state and the error rate in the resting state. It can be seen from Figure 7 that the probability that the error rate is less than 0.2 in the driving state is 93%. The probability that the error rate is less than 0.2 in the rest state is 87%. It can be seen that the error is within the allowable range (<85%), and the stroke division is ideal.

Number all valid travel trajectories to get the ID corresponding to all data points of each travel.

2. Determination of research route based on similarity measurement:

Calculate the similarity between the two trajectories, and finally obtain the similarity matrix of all trajectories. Carry out hierarchical cluster analysis according to the degree of similarity of trajectories.

The trajectory cluster with the highest similarity is determined as the research route. All trajectories in this cluster are used as the basic data source for evaluating the driving behavior of truck drivers; at the same time, all trajectories not on the research route are removed, as shown in Figure 8. Figure 8(a) shows all the travel trajectories of the current vehicle, the number of trajectories is 76; Figure 8(b) shows the travel trajectories of the current vehicle research route, the number of trajectories is 48.

3. Evaluation of driving behavior:

Calculate the frequency at which the operating speed is higher than (vh) speed limit value and lower than (vl) speed limit value, acceleration (a+) and deceleration (a-) exceeding 3.4m/s2 (maximum deceleration threshold) frequency, and The average line break is SD/SG. And standardize the different indicators respectively so that their values are all within the interval [0, 1]. Finally, according to the contribution weights of different indicators, a weighted average total score is obtained, and the higher the value, the greater the potential risk. The driving behavior evaluation of four large trucks is given in this embodiment. The calculation results and the final ranking of potential risk driving are shown in the table below.

Among the four large trucks, the large truck numbered 76 has higher potential risk driving behavior, followed by the large truck numbered 3, and the large truck numbered 90 has very stable and stable driving behavior, with the least potential risk.

Figure 9 shows the actual trajectories and speed change patterns of four large trucks mainly running on highways. It can be seen that the speed range of the large truck No. 76 is very wide, the highest speed is 90 km/h, and the lowest speed is lower than 40 km/h. The speed changes frequently. On the contrary, the speed of the large truck numbered 90 is very uniform, which means it runs very smoothly without frequent shifting behavior. The relative speed of the large truck numbered 3 also varies greatly. Most of the journeys of this large truck maintain a speed between 40km/h and 60km/h, but there are several journeys (same route) below 20km/h. The truck numbered 30 shows relatively low speed on certain trajectories, and the speed distribution shows obvious spatial and regional correlation. A reasonable guess is that the truck may have experienced frequent traffic on these road sections. Congestion, but this also indirectly affects the speed change of the large truck.

The foregoing embodiment is a classic case of the present invention, but the implementation of the present invention is not limited by the foregoing embodiment. Any other changes, modifications, substitutions, combinations, simplifications, etc. that do not deviate from the spirit and principle of the present invention should be equivalent replacement methods and are all included in the protection scope of the present invention.

The present invention can provide reference basis for the following types of industry organizations: (1) It is helpful to test and screen drivers, especially in industries that require high stability and reliability of driving behavior, such as large trucks, buses, and long-distance buses. , School bus and other drivers. (2) Help insurance companies evaluate the benefits of insured drivers. (3) It is helpful for law enforcement agencies to evaluate the potential risk driving of certain drivers, and to request retraining and education of drivers to drive safely.

This application is described with reference to flowcharts and/or block diagrams of methods, equipment (systems), and computer program products according to the embodiments of this application. It should be understood that each process and/or block in the flowchart and/or block diagram, and the combination of processes and/or blocks in the flowchart and/or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing equipment to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing equipment can be generated It is a device that realizes the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device. The device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment. The instructions provide steps for implementing functions specified in a flow or multiple flows in the flowchart and/or a block or multiple blocks in the block diagram.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. Those of ordinary skill in the art can still modify or equivalently replace the specific embodiments of the present invention with reference to the above embodiments. Any modification or equivalent replacement that deviates from the spirit and scope of the present invention shall fall within the protection scope of the claims of the present invention pending approval.

Claims

A method for evaluating the driving behavior of a large truck based on GPS trajectory data is characterized in that it comprises the following steps:

Filter all the driving track data of the vehicle to be tested from the original GPS data;

Process the driving track data to obtain effective driving track data;

Analyze and process the effective driving trajectory data to obtain an effective research trajectory;

Calculate the risky driving behavior index of the effective research trajectory to evaluate the driving behavior.
The method according to claim 1, wherein said deleting all the driving track data of the vehicle to be tested from the original GPS data is specifically based on the number of the vehicle to be tested and the GPS installed on the vehicle to be tested The device number extracts all driving track data of the vehicle to be tested.
The method according to claim 1, wherein the processing the driving trajectory data to obtain valid driving trajectory data; specifically comprising:

Preprocessing the driving track data to obtain the first processed data;

Perform segmentation processing on the first processed data to determine effective travel trajectory data.
The method according to claim 3, wherein the preprocessing of the driving trajectory data to obtain the first processed data specifically includes: clearing abnormal data and duplicate data in the driving trajectory data.
The method according to claim 3, wherein the segmentation processing on the first processed data to determine the effective travel trajectory data specifically comprises:

Filter the first processed data by band-pass filtering algorithm and filter the speed duration and speed interval time;

Compare the speed duration and speed interval with the preset minimum effective stroke length threshold;

The trajectory data corresponding to the speed duration and the speed interval that are greater than the minimum length threshold of the effective stroke are extracted to obtain the effective stroke trajectory data.
The method according to claim 5, wherein the filtering of the first processed data through a band-pass filtering algorithm and filtering the speed duration and the speed interval; specifically comprising:

Determine the error value of the first processed data through a band-pass filtering algorithm, and remove the error value to form the second processed data;

Filter the second processed data to determine the speed duration and the speed interval, and determine whether there is a data gap;

If there is a data gap, judge the data gap time, if the data gap time reaches the set threshold, mark it as a candidate track;

Otherwise, the average interpolation method is used for data restoration.
The method according to claim 6, wherein the screening of the second processed data determines the speed duration and the speed interval, and determines whether there is a data gap; specifically:

By comparing the recorded interval time and calculating the distance between the data points before and after the interval and the average speed quotient, it is determined whether there is a data interval caused by signal loss in the recording interval.
The method according to claim 1, wherein the analyzing and processing the effective driving trajectory data to obtain an effective research trajectory; specifically comprising:

Perform similarity measurement and cluster analysis on effective travel trajectory data;

Match all valid travel trajectories according to their latitude and longitude coordinates;

Determine the route with more repetitions as an effective research trajectory.
The method according to claim 8, wherein the performing similarity measurement and cluster analysis on the effective travel trajectory data specifically comprises:

Choose two travel trajectories from the effective travel trajectory data, and randomly extract some data points for the two trajectories to calculate the similarity of the two travel trajectories;

Calculate the similarity between all trajectories in the effective travel trajectory data, and finally obtain the similarity matrix of all trajectories;

Calculate the variance of clusters between different trajectories to determine the optimal number of clusters.
The method according to claim 1, wherein the calculation of the risky driving behavior index of the effective research trajectory to evaluate the driving behavior specifically comprises:

According to the driving trajectory data corresponding to the optimal number of clusters and the increase in driving load, the driving behavior is evaluated.
The method according to claim 10, wherein the evaluating the driving behavior according to the driving trajectory data corresponding to the optimal number of clusters and the increase of the driving load is specifically:

Determine risk driving indicators; including the frequency of speeding and the frequency of sudden acceleration and deceleration;

Determine the driving load index; this index uses the break rate to calculate the break rate of the vehicle to be tested;

Combining the above two indicators, the risk driving of a single driver is scored; the calculation formula is as follows:

Among them, Fp i is the comprehensive score of the driving behavior trajectory of a large truck driver i; β(n) is the weight coefficient of different indicators;
Is the frequency at which the instantaneous speed of the vehicle under test exceeds the preset safety threshold,
Is the frequency at which the acceleration of the vehicle under test exceeds the preset safety threshold; SD/SG is the break rate, which is the ratio of the total driving time to the total stop time.
According to the method of any one of claims 1-11, a device for evaluating the driving behavior of a large truck based on GPS trajectory data is proposed, which is characterized in that it comprises:

The data screening unit is used to screen all the driving track data of the vehicle to be tested from the original GPS data;

The first processing unit is configured to process the driving trajectory data to obtain effective driving trajectory data;

The second processing unit is used to analyze and process the effective driving trajectory data to obtain an effective research trajectory;

The driving behavior evaluation unit is used to calculate the risk driving behavior index of the effective research trajectory to evaluate the driving behavior.