CN110727706A - Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test - Google Patents
Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test Download PDFInfo
- Publication number
- CN110727706A CN110727706A CN201910820969.6A CN201910820969A CN110727706A CN 110727706 A CN110727706 A CN 110727706A CN 201910820969 A CN201910820969 A CN 201910820969A CN 110727706 A CN110727706 A CN 110727706A
- Authority
- CN
- China
- Prior art keywords
- driving
- risk
- driving scene
- intelligent networked
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Fuzzy Systems (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method for rapidly extracting and grading a risk driving scene for intelligent networking automobile test, which comprises the following steps: (1) normalizing the intelligent networked automobile driving time series data; (2) the intelligent networked automobile driving time sequence data is subjected to dimensionality reduction; (3) the intelligent networked automobile driving time sequence data is subjected to discretization and symbolization processing; (4) and (3) rapidly extracting and carrying out grading processing on the intelligent networked automobile risk driving scene. The method for rapidly extracting and grading the risk driving scenes for the intelligent networked automobile test can rapidly extract and grade various risk driving scenes in the testing working condition of the intelligent networked automobile, and can provide theoretical method support for comprehensive construction and refining optimization of a driving scene library in the subsequent intelligent networked automobile testing stage.
Description
Technical Field
The invention belongs to the field of intelligent automobile testing, and relates to a method for rapidly extracting and grading a risk driving scene for intelligent networking automobile testing.
Background
Research, development, test and road getting are three stages of intelligent networked automobile marketization, and currently, the technical development of the intelligent networked automobiles in China is mainly focused on the test stage. The tests can be classified into software-in-loop, hardware-in-loop, vehicle-in-loop, test field, road tests, and the like. The driving data in the testing stage can be fed back to the research and development stage for product optimization iteration, and the data result of the drive test can serve the testing stage to enrich the capacity of the real driving scene library and the virtual driving scene library, so that the real driving environment can be better restored by the testing field. Throughout the testing phase, driving data is an important source to drive the above process.
The continuous improvement of the vehicle intelligence level and the rapid development of the wireless communication technology provide powerful technical support for the construction of the intelligent networking automobile test platform. At present, vehicle test data acquisition means become extremely abundant, for example, vehicle-mounted real-time data are acquired through a CAN bus technology, vehicle motion information is monitored by a high-precision sensor or an inertial navigation system, and corresponding monitoring pictures are synchronously recorded by a high-definition camera. The type and the quantity of the test data far reach the technical requirements of analysis, but the coding analysis means of the test data does not break through with the increase of the data volume, which brings certain technical obstacles to the subsequent data-driven vehicle test analysis work. On the other hand, for the intelligent internet automobile test, the risky driving scene is an important test segment needing attention, for example, the occurrence of a global first Tesla car accident is caused by insufficient test analysis on the risky driving scene in the automatic driving test stage. The method is very dependent on the technical level of driving data coding, and the research and development of the technical method are limited at present.
Therefore, the method for rapidly extracting and grading the risk driving scene has very important function and significance for the test work of the intelligent networked automobile. In other words, the rapid extraction and classification of the risk driving scene is a core part of the intelligent networked automobile test development, and the effectiveness and comprehensiveness of the intelligent networked automobile test result can be ensured only by accurately, rapidly and effectively extracting and classifying the risk driving events.
Disclosure of Invention
The invention aims to provide an improved method for rapidly extracting and grading a risk driving scene for intelligent networking automobile testing, aiming at the problems in the prior art.
In order to achieve the purpose, the invention adopts the technical scheme that:
a method for rapidly extracting and grading a risk driving scene for intelligent networking automobile testing comprises the following steps:
(1) normalizing the intelligent networked automobile driving time series data;
(2) the intelligent networked automobile driving time sequence data is subjected to dimensionality reduction;
(3) the intelligent networked automobile driving time sequence data is subjected to discretization and symbolization processing;
(4) and (3) rapidly extracting and carrying out grading processing on the intelligent networked automobile risk driving scene.
Preferably, in the step (1), when the raw driving time series value is normalized, the raw driving time series is converted into a normalized time series with a mean value of 0 and a standard deviation of 1.
Preferably, in the step (2), the normalized time series is subjected to dimension reduction processing based on a segment cumulative approximation method.
Furthermore, the driving time sequence data after the dimension reduction processing in the step (2) is changed into a plurality of segmented broken line segments from the initial continuous curve, and the segmented broken line segments in a certain interval are used for replacing the curve segments in the initial interval.
Preferably, in step (3), the time-series data after the dimension reduction processing is subjected to a K-S test, and if the value of the driving time-series data P is greater than 0.05, the driving time-series data (speed, acceleration) after the dimension reduction processing by the segment accumulation approximation method is considered to approximately follow a normal distribution.
And further, dividing the time sequence data subjected to the dimensionality reduction into a plurality of equal probability intervals, calculating specific values of break points among the divided intervals according to a standard normal distribution table, and coding the time sequence values in the same probability interval by using the same symbol to obtain a corresponding symbol sequence.
Preferably, the step (4) specifically comprises the following steps:
a. constructing and dividing risk driving scene working conditions;
b. performing symbolic coding extraction on each type of risk driving scene working condition;
c. paraphrasing the working condition of each type of risk driving scene;
d. the level of the risky driving scenario is graded.
Further, the grades of the risky driving scenario include three grades of high risk, medium risk and low risk.
Preferably, the time-series data comprises at least velocity-time-series data and acceleration-time-series data.
Further, the time series data further comprises one or more of latitude-longitude-time, heading angle-time, triaxial acceleration-time, and CAN data-time series data.
Due to the application of the technical scheme, compared with the prior art, the invention has the following advantages: the invention provides a method for rapidly extracting and grading a risk driving scene for an intelligent networked automobile test, and provides the method for rapidly extracting and grading the risk driving scene for the intelligent networked automobile test. And performing normalization processing, dimension reduction processing, dispersion and symbolization processing on the driving time sequence data in sequence based on a symbolization aggregation approximation method. After semantic coding, the previous dimension is very high, the driving time sequence data with redundant data features is reasonably converted into a symbolic sequence which is strong in readability and easy to search and locate, and the main features of the time sequence data are timely reserved while the data dimension is greatly reduced. On the basis of driving data space-time semantic coding, a risk driving scene definition rule is formulated, symbolic coding of the risk driving scene is achieved, various risk driving scenes in the intelligent internet automobile testing working condition are searched and classified rapidly through character strings, and grading is conducted. The method has obvious advantages and effects on intelligent networked automobile testing, can provide technical support for rapid extraction and grading of risk driving events, and provides theoretical method support for comprehensive construction and refining optimization of a driving scene library in a subsequent intelligent networked automobile testing stage.
Drawings
FIG. 1 is a flow chart of an implementation of a method for rapidly extracting and grading a risky driving scene for intelligent networked automobile testing according to the present invention;
fig. 2(a) and 2(b) are driving time-series data after normalization processing;
fig. 3(a) and 3(b) are driving time-series data after the dimension reduction processing;
fig. 4(a) and 4(b) are driving time-series data after the discretization and symbolization processing;
fig. 5 is a flow chart of a process for rapid extraction and grading of a risky driving scenario.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
The time sequence is a set formed by a group of event data or sequence values based on time follow-up, the follow-up characteristic of the data attribute values on a time axis is reflected, and the time sequence data is large in size, high in dimensionality and high in updating frequency generally. The intelligent networked automobile can acquire massive driving time sequence data in a testing stage, wherein the driving time sequence data comprises various information such as speed, acceleration, steering wheel turning angle, accelerator pedal, brake pedal, video monitoring and the like, and the driving time sequence data simplifying method can greatly reduce the data capacity and provides powerful technical support for rapid extraction and grading of a risk driving scene. The symbolic representation is an effective discretization time sequence dimension reduction method, and has been widely applied in many fields due to the advantages of flexible algorithm, easy operation and the like, but is rarely applied at home and abroad in the field of automobile transportation. The method sequentially performs normalization processing, dimension reduction processing, dispersion and symbolization processing on the intelligent networked automobile driving time sequence data based on symbolized aggregation approximation, and realizes coding of the driving time sequence data. The flow chart of the method implementation is shown in fig. 1, and specifically comprises the following steps:
(1) intelligent networking automobile driving time series data normalization processing
The first step of the method is to normalize the initial driving time series T1,.. and tn, i.e., convert the initial driving time series to a normalized time series with a mean value of 0 and a standard deviation of 1, denoted as T ' ═ T ' 1.., and T ' n. The purpose of this is to eliminate the dimension influence and the influence of the variable self-variation size and the numerical value size, and to facilitate the subsequent dimension reduction and equal probability division of the time series data into several sections, and further to perform the mark encoding of the symbol sequence, where fig. 2 is a certain driving time series data after normalization, fig. 2(a) is a speed time series data, and fig. 2(b) is an acceleration time series data.
(2) Intelligent networking automobile driving time series data dimension reduction processing
The second step of the method is to reduce the dimension of the normalized time series T ' ═ T ' 1.., T ' n based on a segment accumulation approximation method. The normalized time series with length N is converted into a group of time series vectors with length N (N < N)The time series data can be further dispersed and symbolized through reasonable dimension reduction processing. The driving time sequence data are captured, taking speed and acceleration data as examples, the length of each driving time sequence data is A, one driving time sequence data is obtained every second, the total length of the driving time sequence data is A/60min, and n is equal to A. And selecting the dimension N required to be reduced, and respectively carrying out space-time semantic coding on the speed and acceleration time sequence data. For the speed time-series data, the section of the travel speed of the vehicle is about [0, B ] according to its own range]The unit is km/h, and the Nv can be set as C (less than A); for the longitudinal acceleration time series data, corresponding parameters can be set according to the data characteristics, and the intercepted longitudinal acceleration is addedThe speed range is about [ -D1, D2 ]]In the unit of m/s2Na may be set to E (< A). After the parameters are selected, dimension reduction processing can be performed on the captured speed time-series data and acceleration time-series data after normalization processing by a segment accumulation approximation method, for example, fig. 3 shows driving time-series data after dimension reduction processing, fig. 3(a) shows speed time-series data, and fig. 3(b) shows acceleration time-series data.
The driving time sequence data after the dimensionality reduction processing is changed into a plurality of segmented broken line segments from an initial continuous curve, and the segmented broken line segments in a certain interval can replace curve segments in the initial interval, so that the main characteristics of data in the initial interval are kept, the requirement of dimensionality reduction is realized, and the problem of data mining caused by overhigh dimensionality can be avoided.
(3) Intelligent networked automobile driving time sequence data discrete and symbolized processing
The third step of the method is to carry out data discretization on the segmented broken line segment subjected to the dimensionality reduction treatment on the basis of the dimensionality reduction treatment of the second step of the data, and then encode the discretized data according to semantic features (reflecting the semantic features on the speed and the acceleration, namely the value range). And performing K-S test on the time series data (speed and acceleration) after dimension reduction, and if the P value of the driving time series data is more than 0.05, considering that the driving time series data (speed and acceleration) after the dimension reduction by a segmented accumulation approximation method approximately obeys normal distribution. Dividing the time sequence data after dimensionality reduction into m equal probability intervals, wherein break points among the division areas can be obtained through query of a table 1, and beta 1, beta 2. If m is E, the divided E space probabilities are all 1/E, and the values of the division points can be obtained by inquiring a standard normal distribution table. And then, the time sequence values in the same probability interval are coded by the same symbol to obtain a corresponding symbol sequence, so that the discretization and symbolization of the driving time sequence data are completed.
Table 1 breakpoint for number of characters m divided from 2 to 9
The number m of the divided characters needs to be reasonably set according to the scene condition in practical application, and is generally not more than 10 according to experience and application effect. For example, the vehicle speed of the sample data may vary in a range of about [30, 100], the acceleration may vary in a range of about [ -2.5, 2.5], the vehicle speed may be divided by the quotient of its range and 10km/h, and the acceleration may be divided by the quotient of its range and 1m/s2, i.e., ma is 70/10 is 7, mv is 5/1 is 5, and any value near 7 or 5 may be selected. Finally, ma is 7, mv is 4, that is, the vehicle speed and the acceleration are divided into signs by using 7 signs and 4 signs, respectively, and these signs are a, b, c, d, e, f, and g. As shown in fig. 4, the driving time-series data after the discretization and symbolization processing is shown, and fig. 4(a) is the speed time-series data, and fig. 4(b) is the acceleration time-series data.
The driving time series data after the dimension reduction processing, i.e., the piecewise broken line segments (thick lines) are equally probabilistically divided (solid lines) into several partitions, and the determination of the divided partitions can be determined as described above by table 1. After dimension reduction processing, the vehicle speed time sequence data are divided into 7 sections (a, b, c, d, e, f, g) by equal probability, the acceleration time sequence data are divided into 4 sections (a, b, c, d) by equal probability, and the driving time sequence data are subjected to value taking for the symbolic sections as shown in table 2.
TABLE 2 symbolized string intervals of driving time series
Corresponding symbols can be adopted for representing the time sequence in the corresponding division areas, so that the time sequence semantic meaning with high original dimensionality and complicated data characteristics is coded into a visual symbolic sequence, the main characteristics of time sequence data are reserved while the data dimensionality is reduced, the data storage capacity is greatly reduced, and the method has an important effect on rapid extraction of subsequent risk driving scenes. Table 3 shows an example of semantic coding of a certain piece of driving data in the truncated example.
TABLE 3 driving data semantic coding example
(4) Intelligent networked automobile driving time sequence data coding post-processing secondary development
The type of driving data is not limited in the actual encoding process to only the vehicle speed, longitudinal acceleration mentioned in the examples, but these two quantities are chosen in view of the conventional analysis of risky driving scenarios that can usually be done with these two types of data. During secondary development of codes, expansion is carried out according to the requirements of analysis contents, such as the coding work of increasing longitude and latitude, course angle, triaxial acceleration and other data of CAN data (throttle opening, brake signals and the like).
In the above steps (1) - (4), the values of the parameters need to be determined when the driving data is encoded, the values of the parameters are not strictly specified, and the values can be flexibly determined according to the data type characteristics during specific encoding, but still follow a certain principle. The length n of the initial sequence can be selected to be longer, for example, a long-distance transportation task in 8h can be selected, the time frequency can be flexibly determined according to the requirement of analysis precision, the frequency of the speed and acceleration parameters of 1Hz can basically meet the routine analysis, and the steering wheel corner parameters can require 10Hz or higher frequency to achieve an expected analysis effect; the time sequence length N after the dimensionality reduction treatment is basically much smaller than the initial sequence length N, and the specific value can take the value range of the data as a reference; the divided character number m can take the quotient value of the data value range and the precision as a reference basis. In the embodiment, for the sake of concise construction and demonstration of the semantic coding method of the driving data, the adopted example time is short (10min), and the actual process is as described above, a longer whole time period can be selected, and is not completely limited to the example.
Through the coding method in the steps (1) to (4), full-automatic program operation can be realized, semantic coding of data can be rapidly completed only by reading driving data to be coded, and symbolic data to be analyzed can be directly called and sorted. The core essence of the coding method is to convert a numerical time sequence into a symbolic time sequence, so that the coding method brings visual and convenient requirements for analysis. From several steps of the coding method, the coding method has strong reliability, and more evaluation on result validity can be obtained from the application effect of the actual case.
(5) Rapid extraction and grading processing of risk driving scene
The method has the advantages that the method has an important effect on intelligent networked automobile testing by accurately capturing the risk driving scene, the workload of the traditional method for discriminating by means of data threshold is large, and an effective rapid extraction method is provided on the basis of the space-time semantic coding of driving data.
The longitudinal acceleration of the common small passenger car is generally not more than 0.5 g; there are also studies on longitudinal deceleration that use instantaneous deceleration greater than 0.4g as a trigger for a dangerous event. According to the acceleration statistical distribution, the acceleration distribution interval is mainly concentrated between (-0.4g, 0.5g), and the acceleration distribution outside the interval is always a small proportion, because the vehicle is in a stable, medium-low risk driving situation in most time periods when running. Meanwhile, the driver has features such as angry driving and aggressive driving due to the driving style of the driver, or the driver can suddenly accelerate and decelerate for urgent risk avoidance during driving, and the condition is reflected on the vehicle feature parameters, namely, the driver has larger longitudinal acceleration or deceleration. Further, if such a situation occurs at a high vehicle speed, such as a rapid acceleration or a rapid deceleration, a higher risk driving scenario may occur, such as a rapid deceleration in case of an emergency in front of the target vehicle, or a rapid acceleration in case of danger avoidance of the target vehicle, etc., where we define such a situation as a high risk driving event, i.e., a situation where a large acceleration value (absolute value) occurs when the vehicle is traveling at a high vehicle speed.
Specifically, in the intelligent networking automobile testing process, a lot of parameters are used for constructing and describing the risk driving scene working condition, the parameters are selected in practice according to a test target and content, and the two parameters of speed and acceleration are adopted to construct the risk driving scene working condition. After the two parameters are selected, the threshold values of the parameters need to be determined to further divide the operating conditions of the risky driving scene, wherein the speed is divided into three ranges, and the acceleration is divided into four ranges, so that the total number of the operating conditions is 3 × 4 to 12. Then, performing symbolic coding extraction on each type of risk driving scene working condition by using the methods in the steps (1) to (4), and performing corresponding scene paraphrasing. The definition of the risk driving scene grade is based on subjective evaluation and assisted by objective evaluation, and the risk grade is graded by using expert experience on the basis of the definition of the risk driving scene and referring to a popular subjective evaluation method applied in the field of automobile design and research. And finally, completing the one-to-one correspondence of the working condition of the risk driving scene, the extraction of the risk scene, the paraphrasing of the risk scene and the risk classification, and realizing the rapid extraction and classification processing of the risk driving scene by utilizing the coding result, wherein the flow chart is shown in FIG. 5. Specifically, the rapid extraction and grading processing flow of the risky driving scene is as follows:
TABLE 4 Rapid extraction and grading processing method for risk driving scene
And the rapid extraction and classification of various risky driving scenes can be completed by utilizing the table. According to the symbolized coding result, the risk scenes can be represented as ae, af, ag, be, bf, bg, ce, cf, cg, de, df and dg through character strings, so that a more complex risk driving event from the aspects of intuition, qualification and objective quantification is represented as two character strings through symbolization, and various risk driving scenes in the intelligent internet automobile testing working condition can be rapidly extracted and classified through the search of the character strings.
It should be noted that the risky driving scene working condition shown in the invention is described by only adopting two parameters of the vehicle speed and the acceleration, but in the specific intelligent networked automobile testing process, the risky driving scene working condition is often required to be expanded in a targeted manner according to the tested target and content, such as the addition of longitude and latitude, the course angle, the three-axis acceleration, the CAN data (the opening degree of a throttle, a brake signal and the like), the headway, the weather condition, the road type, the information (the age and the driving age) of a driver, the physiological information (the skin electricity, the electrocardio and the electroencephalogram) of the driver and the like.
The invention provides a method for rapidly extracting and grading a risk driving scene aiming at intelligent networking automobile testing. And performing normalization processing, dimension reduction processing, dispersion and symbolization processing on the driving time sequence data in sequence based on a symbolization aggregation approximation method. After semantic coding, the previous dimension is very high, the driving time sequence data with redundant data features is reasonably converted into a symbolic sequence which is strong in readability and easy to search and locate, and the main features of the time sequence data are timely reserved while the data dimension is greatly reduced. On the basis of the driving data space-time semantic coding, a risk driving scene defining rule is formulated, symbolic coding of the risk driving scene is achieved, and various risk driving scenes in the intelligent internet automobile testing working condition are extracted and classified through rapid search of character strings. The method has obvious advantages and effects on intelligent networked automobile testing, can provide technical support for rapid extraction and grading of risk driving events, and provides theoretical method support for comprehensive construction and refining optimization of a driving scene library in a subsequent intelligent networked automobile testing stage.
The above-mentioned embodiments are merely illustrative of the technical idea and features of the present invention, and the purpose thereof is to enable those skilled in the art to understand the contents of the present invention and implement the present invention, and not to limit the scope of the present invention, and all equivalent changes or modifications made according to the spirit of the present invention should be covered in the scope of the present invention.
Claims (10)
1. A method for rapidly extracting and grading a risk driving scene for intelligent networking automobile test is characterized by comprising the following steps: the method comprises the following steps:
(1) normalizing the intelligent networked automobile driving time series data;
(2) the intelligent networked automobile driving time sequence data is subjected to dimensionality reduction;
(3) the intelligent networked automobile driving time sequence data is subjected to discretization and symbolization processing;
(4) and (3) rapidly extracting and carrying out grading processing on the intelligent networked automobile risk driving scene.
2. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 1, wherein the method comprises the following steps: in the step (1), when the values of the original driving time series are normalized, the original driving time series are converted into a normalized time series with a mean value of 0 and a standard deviation of 1.
3. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 1, wherein the method comprises the following steps: in the step (2), dimension reduction processing is carried out on the time series after the standardized normalization processing based on a segmented cumulative approximation method.
4. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method according to claim 1 or 3, characterized in that: and (3) changing the driving time sequence data subjected to the dimension reduction processing in the step (2) from an initial continuous curve into a plurality of segmented broken line segments, and replacing the initial curve segment in a certain interval by the segmented broken line segment in the interval.
5. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 1, wherein the method comprises the following steps: in the step (3), the time series data after the dimensionality reduction is subjected to K-S test, and if the P value of the driving time series data is more than 0.05, the driving time series data (speed and acceleration) after the dimensionality reduction by the segmented cumulative approximation method is considered to approximately follow normal distribution.
6. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 5, wherein the method comprises the following steps: dividing the time sequence data after the dimensionality reduction processing into a plurality of equal probability intervals, calculating specific values of break points among the divided intervals according to a standard normal distribution table, and coding the time sequence values in the same probability interval by using the same symbol to obtain a corresponding symbol sequence.
7. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 1, wherein the method comprises the following steps: the step (4) specifically comprises the following steps:
a. constructing and dividing risk driving scene working conditions;
b. performing symbolic coding extraction on each type of risk driving scene working condition;
c. paraphrasing the working condition of each type of risk driving scene;
d. the level of the risky driving scenario is graded.
8. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 7, wherein the method comprises the following steps: the grades of the risky driving scene comprise three grades of high risk, medium risk and low risk.
9. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 1, wherein the method comprises the following steps: the time series data includes at least velocity-time series data and acceleration-time series data.
10. The intelligent networked automobile test-oriented risk driving scene rapid extraction and classification method as claimed in claim 9, wherein the method comprises the following steps: the time series data further includes one or more of latitude-longitude-time, heading angle-time, triaxial acceleration-time, and CAN data-time series data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910820969.6A CN110727706A (en) | 2019-09-02 | 2019-09-02 | Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910820969.6A CN110727706A (en) | 2019-09-02 | 2019-09-02 | Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110727706A true CN110727706A (en) | 2020-01-24 |
Family
ID=69218750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910820969.6A Pending CN110727706A (en) | 2019-09-02 | 2019-09-02 | Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110727706A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401414A (en) * | 2020-02-29 | 2020-07-10 | 同济大学 | Natural driving data-based dangerous scene extraction and classification method |
| CN111735639A (en) * | 2020-05-26 | 2020-10-02 | 清华大学苏州汽车研究院（相城） | Automatic driving scene minimum set generation method for intelligent networked automobile demonstration area |
| CN112765812A (en) * | 2021-01-19 | 2021-05-07 | 中国科学院软件研究所 | Autonomous ability rapid evaluation method and system for unmanned system decision strategy |
| WO2022110978A1 (en) * | 2020-11-24 | 2022-06-02 | Suzhou Zhijia Science & Technologies Co., Ltd. | Method and apparatus for driving data search |
| CN115249408A (en) * | 2022-06-21 | 2022-10-28 | 重庆长安汽车股份有限公司 | Scene classification extraction method for automatic driving test data |
| CN116246468A (en) * | 2023-03-14 | 2023-06-09 | 西安科技大学 | Multi-element space-time data-based distracted driving risk road section identification and control method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150344030A1 (en) * | 2014-05-30 | 2015-12-03 | Honda Research Institute Europe Gmbh | Method and vehicle with an advanced driver assistance system for risk-based traffic scene analysis |
| CN105374211A (en) * | 2015-12-09 | 2016-03-02 | 敏驰信息科技（上海）有限公司 | System and method for calculating driving risks and assisting automobile insurance pricing based on multi-source data |
| CN108332977A (en) * | 2018-01-23 | 2018-07-27 | 常熟昆仑智能科技有限公司 | A kind of classifying and analyzing method joining automotive test scene to intelligent network |
-
2019
- 2019-09-02 CN CN201910820969.6A patent/CN110727706A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150344030A1 (en) * | 2014-05-30 | 2015-12-03 | Honda Research Institute Europe Gmbh | Method and vehicle with an advanced driver assistance system for risk-based traffic scene analysis |
| CN105374211A (en) * | 2015-12-09 | 2016-03-02 | 敏驰信息科技（上海）有限公司 | System and method for calculating driving risks and assisting automobile insurance pricing based on multi-source data |
| CN108332977A (en) * | 2018-01-23 | 2018-07-27 | 常熟昆仑智能科技有限公司 | A kind of classifying and analyzing method joining automotive test scene to intelligent network |
Non-Patent Citations (1)
| Title |
|---|
| 孙川 等: "《基于 SAX 的车载数据时空语义编码及分析方法》" * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111401414A (en) * | 2020-02-29 | 2020-07-10 | 同济大学 | Natural driving data-based dangerous scene extraction and classification method |
| CN111401414B (en) * | 2020-02-29 | 2023-02-10 | 同济大学 | Natural driving data-based dangerous scene extraction and classification method |
| CN111735639A (en) * | 2020-05-26 | 2020-10-02 | 清华大学苏州汽车研究院（相城） | Automatic driving scene minimum set generation method for intelligent networked automobile demonstration area |
| CN111735639B (en) * | 2020-05-26 | 2022-03-22 | 清华大学苏州汽车研究院（相城） | Automatic driving scene minimum set generation method for intelligent networked automobile demonstration area |
| WO2022110978A1 (en) * | 2020-11-24 | 2022-06-02 | Suzhou Zhijia Science & Technologies Co., Ltd. | Method and apparatus for driving data search |
| CN112765812A (en) * | 2021-01-19 | 2021-05-07 | 中国科学院软件研究所 | Autonomous ability rapid evaluation method and system for unmanned system decision strategy |
| CN112765812B (en) * | 2021-01-19 | 2021-09-07 | 中国科学院软件研究所 | Autonomous ability rapid evaluation method and system for unmanned system decision strategy |
| CN115249408A (en) * | 2022-06-21 | 2022-10-28 | 重庆长安汽车股份有限公司 | Scene classification extraction method for automatic driving test data |
| CN116246468A (en) * | 2023-03-14 | 2023-06-09 | 西安科技大学 | Multi-element space-time data-based distracted driving risk road section identification and control method |
| CN116246468B (en) * | 2023-03-14 | 2023-11-14 | 西安科技大学 | Multi-element space-time data-based distracted driving risk road section identification and control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110727706A (en) | Method for rapidly extracting and grading risk driving scene for intelligent networking automobile test | |
| US20150352999A1 (en) | Driving context generation system | |
| CN113609016B (en) | Method, device, equipment and medium for constructing automatic driving test scene of vehicle | |
| CN110304068B (en) | Method, device, equipment and storage medium for collecting automobile driving environment information | |
| CN113159105B (en) | Driving behavior unsupervised mode identification method and data acquisition monitoring system | |
| CN106956680A (en) | A kind of electric automobile driving behavior identifying and analyzing method | |
| CN105427620A (en) | Taxi operation data-based illegal-operating vehicle identification method | |
| CN115588310A (en) | Vehicle collision risk prediction method based on trajectory data | |
| Wu et al. | Clustering of several typical behavioral characteristics of commercial vehicle drivers based on GPS data mining: Case study of highways in China | |
| CN110458214A (en) | Driver replaces recognition methods and device | |
| CN115291515A (en) | Automatic driving simulation test system and evaluation method based on digital twinning | |
| Montanari et al. | Pattern recognition for driving scenario detection in real driving data | |
| CN110175218B (en) | Method, device and equipment for generating navigation broadcast content and computer storage medium | |
| CN110097144B (en) | Tire automatic detection big data and cloud computing system and application research thereof | |
| Brahim et al. | A machine learning smartphone-based sensing for driver behavior classification | |
| CN114912689A (en) | Map grid index and XGBOST-based over-limit vehicle destination prediction method and system | |
| CN114428717A (en) | Automatic driving test scene set updating method and system based on real data | |
| US10415981B2 (en) | Anomaly estimation apparatus and display apparatus | |
| CN112269827B (en) | Data processing method and device, computer equipment and computer readable storage medium | |
| Yarlagadda et al. | Heterogeneity in the driver behavior: an exploratory study using real-time driving data | |
| CN116753938A (en) | Vehicle test scene generation method, device, storage medium and equipment | |
| CN116307699A (en) | Road hidden trouble point segment grading method, device and storage medium based on multi-source data | |
| CN115841765A (en) | Vehicle position blind area monitoring method and device, electronic equipment and readable storage medium | |
| CN113033713B (en) | Accident fragment identification method, device, equipment and readable storage medium | |
| Xu et al. | Novel fast safety assessment method for the buffer section of maintenance work zone |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |