CN112232525B - Driving mode characteristic construction and screening method and device and storage medium - Google Patents
Driving mode characteristic construction and screening method and device and storage medium Download PDFInfo
- Publication number
- CN112232525B CN112232525B CN202011471000.1A CN202011471000A CN112232525B CN 112232525 B CN112232525 B CN 112232525B CN 202011471000 A CN202011471000 A CN 202011471000A CN 112232525 B CN112232525 B CN 112232525B
- Authority
- CN
- China
- Prior art keywords
- feature
- features
- driving
- screening
- driving mode
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The invention discloses a driving mode characteristic construction and screening method and device and a computer readable storage medium, wherein the method comprises the following steps: observing driving operation parameters in a driving mode sample by using a synchronous multi-time window method, and constructing a feature space of the driving mode sample by using statistical features and personalized features; performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features; and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample. The method and the device solve the problems of few characteristic feature information and poor feature screening effect of the driving mode in the traditional technology, realize the expansion of the characteristic feature of the driving mode, enhance the feature screening effect and improve the identification precision of the driving mode.
Description
Technical Field
The present application relates to the field of feature engineering technologies, and in particular, to a driving pattern feature construction and screening method and apparatus, and a computer-readable storage medium.
Background
The driving mode mainly refers to a combination of specific driving operation behaviors taken by a driver when completing various driving strategies (such as path selection). Common driving modes include straight driving, lane changing, car following and the like. The driving mode of the driver is closely associated with the current driving road condition, the analysis and identification of the driving mode are beneficial to understanding the behavior of the driver in the driving process, and the characteristics of the driving mode under each danger degree are analyzed, so that the road management is facilitated, the traffic environment is improved, and the traffic safety is improved. In addition, great help is designed for the driving assistance system, and the function and the performance of the driving assistance system can be improved.
The current analysis, identification and prediction of the driving mode mainly adopts a machine learning method, namely a feature addition algorithm, but has certain defects. The first drawback is that the characterization feature information is small. The current characteristic structure of the driving mode is still selected and expanded according to experience, and the characteristics of the driving mode cannot be completely expressed. The second drawback is poor feature screening methods. In the current feature screening, the traditional screening algorithm such as principal component analysis is mainly relied on, and the limitation of the feature screening algorithm is not considered. These defects make the recognition accuracy of the driving pattern unable to be improved all the time, and help for online recognition is very limited.
Disclosure of Invention
By providing the driving mode feature construction and screening method and device and the computer readable storage medium, the problems that in the prior art, the characterization feature information of the driving mode is few and the feature screening effect is poor are solved, the extension of the characterization feature of the driving mode is realized, the feature screening effect is enhanced, and the identification precision of the driving mode is improved.
The embodiment of the application provides a driving mode characteristic construction and screening method, which comprises the following steps:
observing driving operation parameters in a driving mode sample by using a synchronous multi-time window method, and constructing a feature space of the driving mode sample by using statistical features and personalized features; the synchronous multi-time window method is a time sequence data length selection method for backward deducing different time lengths by taking a current time node as a reference;
performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features;
and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample.
In one embodiment, the step of observing the driving operation parameters in the driving pattern sample using the synchronized multi-time window method comprises:
a preset number of time windows are constructed for the driving operational parameters in the driving pattern samples by taking a specific time node as a reference and back-stepping different lengths of time.
In one embodiment, the step of constructing the feature space of the driving pattern sample by using the statistical features and the personalized features comprises:
and constructing a feature space of the driving pattern sample by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample and the personalized features of the driving operation parameters with the typical features in each time window.
In one embodiment, the step of performing a saliency analysis on the features in the feature space and reducing the feature dimension of the feature space by rejecting extraneous features includes:
judging whether the features have significance differences among different driving modes or not by performing significance analysis on the features in the feature space;
and if the characteristic has no significant difference among different driving modes, removing the characteristic from the characteristic space.
In one embodiment, the driving operation parameters in the driving pattern samples include at least an accelerator pedal position, a vehicle speed, a brake pedal position, a steering wheel angle, a lateral acceleration, and a yaw rate.
In one embodiment, the typical characteristic driving operation parameter is a steering wheel angle.
In one embodiment, the feature screening algorithm comprises a sequence forward selection algorithm and a sequence forward floating selection algorithm.
In an embodiment, the feature dimension of the current optimal feature space is less than or equal to a preset value.
The embodiment of the application also provides a driving mode feature constructing and screening device, which comprises a processor, a memory and a feature constructing and screening program which is stored on the memory and can run on the processor, wherein the feature constructing and screening program realizes the steps of the driving mode feature constructing and screening method when being executed by the processor.
The embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores a feature construction and a screening program, and the feature construction and the screening program are executed by a processor to realize the steps of the driving mode feature construction and the screening method.
The technical scheme of the driving mode feature construction and screening method and device and the computer readable storage medium provided by the embodiment of the application has at least the following technical effects:
the driving operation parameters in the driving mode sample are observed by using a synchronous multi-time window method, and the feature space of the driving mode sample is constructed by using statistical features and personalized features; performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features; and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample. Therefore, the problems that the representation characteristic information of the driving mode in the traditional technology is few and the characteristic screening effect is poor are effectively solved, the extension of the representation characteristic of the driving mode is realized, the characteristic screening effect is enhanced, and the identification precision of the driving mode is improved.
Drawings
FIG. 1 is a schematic structural diagram of an apparatus according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart illustrating a first embodiment of a driving pattern feature construction and screening method according to the present application;
FIG. 3 is a schematic flow chart illustrating a second embodiment of a driving pattern feature construction and screening method according to the present application;
FIG. 4 is a schematic flow chart illustrating a third embodiment of a driving pattern feature construction and screening method according to the present application;
fig. 5 is a schematic diagram of a method for synchronizing multiple time windows according to an embodiment of the present application.
Detailed Description
In order to solve the problems of few characteristic feature information and poor feature screening effect of a driving mode in the traditional technology, a synchronous multi-time window method is adopted to observe driving operation parameters in a driving mode sample, and a feature space of the driving mode sample is constructed by utilizing statistical features and personalized features; performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features; and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the technical scheme of the current optimal feature space of the driving mode sample. The method and the device have the advantages that the characteristic features of the driving mode are expanded, the feature screening effect is enhanced, and the identification precision of the driving mode is improved.
For a better understanding of the above technical solutions, exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Referring to fig. 1, it is a schematic diagram of a hardware structure of an apparatus involved in various embodiments of the present application, where the apparatus may include: processor 101, memory 102, input module 103, output module 104, and the like. Those skilled in the art will appreciate that the hardware configuration of the apparatus shown in fig. 1 does not constitute a limitation of the apparatus, which may include more or less components than those shown, or some components in combination, or a different arrangement of components.
The various components of the device are described in detail below with reference to fig. 1:
the processor 101 is a control center of the apparatus, connects various parts of the entire apparatus, and performs various functions of the apparatus or processes data by running or executing a program stored in the memory 102 and calling up the data stored in the memory 102, thereby monitoring the entire apparatus.
The memory 102 may be used to store various programs of the device as well as various data. The memory 102 mainly includes a program storage area and a data storage area, wherein the program storage area at least stores programs required for feature construction and screening; the storage data area may store various data of the device. Further, the memory 102 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.
The input module 103 may be used to input driving pattern samples for feature construction and screening.
The output module 104 may be used to output the results of the feature construction and the results of the feature screening.
In the embodiment of the present application, the processor 101 may be configured to invoke the feature construction and filter program stored in the memory 102, and perform the following operations:
observing driving operation parameters in a driving mode sample by using a synchronous multi-time window method, and constructing a feature space of the driving mode sample by using statistical features and personalized features;
performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features;
and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample.
In one embodiment, the processor 101 may be configured to invoke the feature builder and filter stored in the memory 102 and perform the following operations:
a preset number of time windows are constructed for the driving operational parameters in the driving pattern samples by taking a specific time node as a reference and back-stepping different lengths of time.
In one embodiment, the processor 101 may be configured to invoke the feature builder and filter stored in the memory 102 and perform the following operations:
and constructing a feature space of the driving pattern sample by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample and the personalized features of the driving operation parameters with the typical features in each time window.
In one embodiment, the processor 101 may be configured to invoke the feature builder and filter stored in the memory 102 and perform the following operations:
judging whether the features have significance differences among different driving modes or not by performing significance analysis on the features in the feature space;
and if the characteristic has no significant difference among different driving modes, removing the characteristic from the characteristic space.
In one embodiment, the driving operation parameters in the driving pattern samples include at least an accelerator pedal position, a vehicle speed, a brake pedal position, a steering wheel angle, a lateral acceleration, and a yaw rate.
In one embodiment, the typical characteristic driving operation parameter is a steering wheel angle.
In one embodiment, the feature screening algorithm comprises a sequence forward selection algorithm and a sequence forward floating selection algorithm.
In an embodiment, the feature dimension of the current optimal feature space is less than or equal to a preset value.
According to the technical scheme, the driving operation parameters in the driving mode sample are observed by using a synchronous multi-time window method, and the feature space of the driving mode sample is constructed by using statistical features and personalized features; performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features; and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample. Therefore, the problems that the representation characteristic information of the driving mode in the traditional technology is few and the characteristic screening effect is poor are effectively solved, the extension of the representation characteristic of the driving mode is realized, the characteristic screening effect is enhanced, and the identification precision of the driving mode is improved.
For better understanding of the above technical solutions, the following detailed descriptions will be provided in conjunction with the drawings and the detailed description of the embodiments.
Referring to fig. 2, in a first embodiment of the present application, a driving pattern feature configuration and screening method specifically includes the following steps:
step S110, observing driving operation parameters in a driving mode sample by using a synchronous multi-time window method, and constructing a feature space of the driving mode sample by using statistical features and personalized features; the synchronous multi-time window method is a time sequence data length selection method for backward reversing different time lengths by taking the current time node as a reference.
In the present embodiment, the driving mode mainly refers to a combination of a plurality of specific driving operation behaviors that the driver takes when completing various driving strategies (such as route selection). Common driving modes include straight driving, lane changing, car following and the like. The driving operation parameters are parameters related to driving operation, and may include vehicle speed, accelerator pedal position, and the like. Since the driving mode of the driver is a time dimension behavior and the driving mode is a combination of a plurality of driving operation behaviors, the driving operation parameters in the driving mode sample can be observed by using a synchronous multi-time window method. The synchronous multi-time window method is a time sequence data length selection method for backward deducing different time lengths by taking the current time node as a reference. As shown in fig. 5, in the synchronous multi-time window diagram with the feature value as the vertical axis and the time as the horizontal axis, there are three time windows with different widths (for convenience of distinction, the three time windows have different heights, and the heights have no practical meaning). Wherein the right frames of the three time windows coincide (i.e., the current time nodes coincide). The synchronous multi-time window method can express information contained in data from multiple time dimensions, and time windows with different widths can express different information. Taking the lane change mode in the driving mode as an example, when the width of the time window is 0.2s, the time window describes only the instantaneous dynamic characteristics of the driving operation, and the difference between the lane change mode and other modes is difficult to judge; and when the width of the time window is 3s, the time window describes the time period trend characteristic of the driving operation in the time period, and the time window can transmit judgment information whether the lane change mode is adopted in the current time period. The method and the device have the advantages that the advantages of the long time window and the short time window are comprehensively utilized, the synchronous multi-time window method is adopted to carry out multi-dimensional analysis on the features, and the information expression of the features is optimized by combining the instantaneous dynamic characteristics and the time-interval trend characteristics of the driving operation.
After the driving operation parameters in the driving mode sample are observed by using the synchronous multi-time window method, a certain method is adopted to perform characteristic expansion on the driving mode, and the method adopted by the application is statistical characteristic expansion and personalized characteristic expansion. The statistical characteristic expansion method is used for expanding the characteristics of the driving mode by calculating the statistical characteristics of all driving operation parameters in the driving mode sample; the personalized feature expansion method is a method which is very effective in solving specific problems by taking the driving operation parameters with typical features in the driving mode as expansion standards and constructing some specific features according to the characteristics of the driving operation parameters to expand the features of the driving mode. The feature expression of the driving mode can be greatly enriched through a synchronous multi-time window method, statistical feature expansion and personalized feature expansion, and relatively complete feature information is obtained, so that the feature space of the driving mode sample can be constructed by utilizing the statistical features and the personalized features.
And step S120, performing significance analysis on the features in the feature space, and reducing the feature dimension of the feature space by removing irrelevant features.
In this embodiment, the significance analysis means that an assumption is made on a parameter of the population or a distribution form of the population in advance, and then whether the assumption is reasonable or not is determined by using sample information, that is, whether the true situation of the population is significantly different from the original assumption or not is determined. The significance analysis can be used to determine whether there is a significant difference in certain features between different groups, and in this application, whether there is a significant difference in each feature between different driving patterns (e.g., left lane change, right lane change, straight-ahead). After the features in the feature space are subjected to significance analysis, whether significance differences exist among different driving modes of each feature in the feature space can be known, and then irrelevant features which do not have significance differences among different driving modes are removed from the feature space, so that the primary screening of the features in the feature space can be realized, and the feature dimensionality of the feature space is reduced.
And S130, performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample.
In this embodiment, feature screening refers to selecting a feature combination that can more completely express all information in a constructed feature space. The feature screening can reduce feature dimension, centralize main information, reduce resource consumption during processing and improve calculation efficiency. At present, a plurality of feature screening algorithms are available, but all feature screening algorithms have certain limitations, and feature subsets obtained through the feature screening algorithms are often locally optimal, but not globally optimal. Before the feature screening algorithm is used, the features in the feature space are screened in one round by applying significance analysis, redundant irrelevant features are removed, and the possibility that the feature screening algorithm falls into local optimization is reduced. Therefore, at this time, feature selection is performed on the features in the feature space after the dimension reduction by using a feature screening algorithm, so that the current optimal feature space of the driving pattern sample can be obtained. The feature screening algorithm can be an algorithm for removing irrelevant features and selecting effective features by utilizing the relevance between the features and the labels; or an algorithm for judging whether the features are effective or not by using a posterior method, for example, training by using a certain machine learning model to obtain the scores of the features and then selecting the scores. In summary, the feature screening algorithm needs to determine the correlation between the features and the tags, i.e. whether the features can effectively distinguish the tags. In one embodiment, the feature screening algorithm includes a sequence forward selection algorithm (SFS), a sequence forward floating selection algorithm (SFFS). In addition, the feature dimension of the current optimal feature space is less than or equal to a preset value. The preset value is used for controlling the feature dimension, so that the resource consumption during processing is reduced, and the calculation efficiency is improved. For example, the preset value may be 10.
The method has the advantages that the driving operation parameters in the driving mode sample are observed by using a synchronous multi-time window method, and the feature space of the driving mode sample is constructed by using statistical features and personalized features; performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features; and performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the technical scheme of the current optimal feature space of the driving mode sample. Therefore, the problems that the representation characteristic information of the driving mode in the traditional technology is few and the characteristic screening effect is poor are effectively solved, the extension of the representation characteristic of the driving mode is realized, the characteristic screening effect is enhanced, and the identification precision of the driving mode is improved.
Referring to fig. 3, in a second embodiment of the present application, a driving pattern feature configuration and screening method specifically includes the following steps:
step S211, constructing a preset number of time windows for the driving operation parameters in the driving pattern sample by taking the specific time node as a reference and backward pushing time of different lengths.
In this embodiment, a preset number of time windows may be constructed for each driving operation parameter in the driving pattern sample by backward-pushing a plurality of different lengths of time with reference to a particular time node. The specific time node may be a time node for completing the driving mode operation in the driving mode sample, or may be set as another suitable time node according to actual needs. The different lengths of time may be selected according to the time dimension characteristics of the driving pattern. For example, for a driving mode with a short time span, a short time length may be set; while for a driving mode with a longer time span, a longer time length may be set. And there may be regularity between a plurality of different time lengths, for example, 0.5s, 1s, 1.5 s; there may be no regularity, for example, 0.5s, 1.4s, 2 s. The preset number can be set according to actual needs. Namely, in practical application, the time window width and the time window number can be dynamically adjusted according to the scene task requirements. In one embodiment, the driving operation parameters in the driving pattern samples include accelerator pedal position, vehicle speed, brake pedal position, steering wheel angle, lateral acceleration, yaw rate.
Step S212, a feature space of the driving pattern sample is constructed by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample and the personalized features of the driving operation parameters with the typical features in each time window.
In this embodiment, the statistical characteristics may include 11 statistical characteristics in table 1 below.
TABLE 1
If 6 time windows are constructed in step S211, and the driving operation parameters are the position of the accelerator pedal, the vehicle speed, the position of the brake pedal, the steering wheel angle, the lateral acceleration, and the yaw rate, 396 features of the driving pattern sample can be obtained by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample. Assuming that there are 10 individualized features for the characteristic driving operation parameters, 60 features of the driving pattern sample can be obtained by calculating the individualized features of the characteristic driving operation parameters in each time window. That is, 456 features of the driving pattern sample can be obtained in total, and the feature expression of the driving pattern is greatly enriched. In one embodiment, the typical characteristic driving operation parameter is a steering wheel angle. In some driving modes, the steering wheel angle is obviously provided with a typical characteristic, and the driving mode can be well distinguished from other driving modes.
For the steering wheel turning angle, firstly, the steering wheel turning angle rate can be calculated according to the turning angle values at different time nodes, and then, personalized features are introduced according to the steering wheel turning angle and the steering wheel turning angle rate. Wherein the personalized features may be as shown in table 2 below:
TABLE 2
The corner ellipse index formula is
The distance index formula is
. Wherein the content of the first and second substances,
which indicates the angle of rotation of the steering wheel,
which is indicative of the rate of steering wheel rotation,
is the undetermined coefficient. In practical application, personalized features can be dynamically constructed according to scene task requirements.
Step S220, performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features.
And step S230, performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample.
The method has the beneficial effect that the step of performing feature expansion is refined on the basis of the first embodiment. Therefore, the problems that the representation characteristic information of the driving mode in the traditional technology is few and the characteristic screening effect is poor are further effectively solved, the extension of the representation characteristic of the driving mode is realized, the characteristic screening effect is enhanced, and the identification precision of the driving mode is improved.
Referring to fig. 4, in a third embodiment of the present application, a driving pattern feature configuration and screening method specifically includes the following steps:
in step S311, a preset number of time windows are constructed for the driving operation parameters in the driving pattern sample by backward-pushing time of different lengths with the specific time node as a reference.
Step S312, a feature space of the driving pattern sample is constructed by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample and the personalized features of the driving operation parameters with the typical features in each time window.
Step S321, performing significance analysis on the features in the feature space, and determining whether there is a significant difference between the features in different driving modes.
In this embodiment, the step of performing significance analysis on the features in the feature space may be: firstly, the significant difference of the features in the feature space among different driving modes is sequentially assumed; then using variance test to test the data of the features in different driving modes; if the p value of the variance test is less than or equal to the significance level, the feature is considered to have significance difference among different driving modes; if the p-value of the variance test is greater than the significance level, the feature is considered to be not significantly different between different driving modes. Wherein common values for the significance level include 0.05, 0.01.
Step S322, if the features have no significant difference among different driving modes, the features are removed from the feature space.
In this embodiment, if there is no significant difference between the features in different driving modes, it means that the features cannot be used to effectively distinguish different driving modes, and at this time, the features need to be removed from the feature space, so as to remove redundant extraneous features, thereby reducing the possibility of falling into local optimality when performing feature screening.
And S330, performing feature selection on the features in the feature space after the dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample.
In this embodiment, the process and the result of verifying the technical solution of the present application by using the real data are shown as follows.
The selected driving mode sample comprises a left lane changing sample, a right lane changing sample and a straight sample; the selected driving operation parameters comprise an accelerator pedal position, a vehicle speed, a brake pedal position, a steering wheel turning angle, a transverse acceleration and a yaw angular velocity; the selected time windows are 6 time windows with the widths of 0.5s, 1s, 1.5s, 2s, 2.5s and 3s respectively; the statistical features selected were 11 features in the foregoing table; the selected personalized features are 10 features in the table; the significance level selected was 0.05.
456 features are obtained after the operations of step S311 and step S312; the 456 features result in 296 features after the operations of step S321 and step S322; in step S330, 296 features are processed by using a random feature subset selection algorithm (RSFS), a sequence forward selection algorithm (SFS), and a sequence forward floating selection algorithm (SFFS), respectively, and the results are shown in table 3 below:
TABLE 3
For comparison, without performing the operations of step S321 and step S322, 456 features are processed directly by using a random feature subset selection algorithm (RSFS), a sequence forward selection algorithm (SFS), and a sequence forward floating selection algorithm (SFFS), and the results are shown in table 4 below:
TABLE 4
The validation data set contains three types of samples: 250 left lane samples, 325 right lane samples and 300 straight samples. Each sample is represented by the previously screened features and has a corresponding label (-1: left lane change, 0: straight, 1: right lane change). And dividing each type of sample into a training set and a testing set according to 4: 1. And respectively training Hidden Markov Models (HMMs) for identifying different driving modes by taking the training samples of each type of driving mode as input to obtain an HMM-left lane changing model, an HMM-straight-moving model and an HMM-right lane changing model. And inputting the test samples of the three types of driving modes into the three models, giving the appearance probability (namely the score) of the sample in the current model by the three models, and comparing the maximum value of the probability, wherein the corresponding model class is the final mode class. The final results are shown in table 5 below, where single screen is the signature screening protocol without significance analysis and double screen is the signature screening protocol of the present application.
TABLE 5
From the results, after the significance analysis, only when the random feature subset selection algorithm (RSFS) is used, the final recognition accuracy is reduced compared with that when the significance analysis is not used, and the other two are obviously improved (both are more than 20%). Among them, the significance analysis combined with the sequence forward selection algorithm (SFS) performed best in the current problem, and the dual-screen scheme achieved 82.3% accuracy. It can be seen that the double-screen scheme is very effective for improving the precision, and can make up the limitation of a feature screening algorithm to a certain extent.
The method has the beneficial effect that the step of carrying out preliminary feature screening is refined on the basis of the second embodiment. Therefore, the problems that the representation characteristic information of the driving mode in the traditional technology is few and the characteristic screening effect is poor are further effectively solved, the extension of the representation characteristic of the driving mode is realized, the characteristic screening effect is enhanced, and the identification precision of the driving mode is improved.
Based on the same inventive concept, the embodiment of the present application further provides a driving mode feature constructing and screening apparatus, where the apparatus includes a processor, a memory, and a feature constructing and screening program stored in the memory and capable of running on the processor, and when executed by the processor, the feature constructing and screening program implements the processes of the driving mode feature constructing and screening method embodiments, and can achieve the same technical effects, and is not described herein again to avoid repetition.
Since the driving mode feature configuration and screening device provided in the embodiment of the present application is a device used for implementing the method in the embodiment of the present application, a person skilled in the art can understand the specific structure and the deformation of the device based on the method described in the embodiment of the present application, and thus the detailed description is omitted here. All devices used in the methods of the embodiments of the present application are within the scope of the present application.
Based on the same inventive concept, an embodiment of the present application further provides a computer-readable storage medium, where a feature structure and a screening program are stored on the computer-readable storage medium, and when executed by a processor, the feature structure and the screening program implement the processes of the driving mode feature structure and the screening method embodiment, and can achieve the same technical effects, and are not described herein again to avoid repetition.
Since the computer-readable storage medium provided in the embodiments of the present application is a computer-readable storage medium used for implementing the method in the embodiments of the present application, based on the method described in the embodiments of the present application, those skilled in the art can understand the specific structure and modification of the computer-readable storage medium, and thus details are not described herein. Any computer-readable storage medium that can be used with the methods of the embodiments of the present application is intended to be within the scope of the present application.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the invention without departing from the invention
With clear spirit and scope. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (7)
1. A driving pattern feature construction and screening method, characterized in that the method comprises:
observing driving operation parameters in a driving mode sample by using a synchronous multi-time window method, and constructing a feature space of the driving mode sample by using statistical features and personalized features; the synchronous multi-time window method is a time sequence data length selection method for backward deducing different time lengths by taking a current time node as a reference, and the synchronous multi-time window method is adopted to describe the instantaneous dynamic characteristic and the time period trend characteristic of the driving operation parameter;
performing significance analysis on the features in the feature space, and reducing feature dimensions of the feature space by removing irrelevant features;
performing feature selection on features in the feature space after dimension reduction by using a feature screening algorithm to obtain the current optimal feature space of the driving mode sample; the feature screening algorithm comprises a sequence forward selection algorithm;
wherein the step of observing the driving operation parameters in the driving pattern sample using the synchronized multi-time window method comprises:
establishing a preset number of time windows for the driving operation parameters in the driving mode sample by taking the specific time node as a reference and backwards pushing time with different lengths;
the step of constructing the feature space of the driving pattern sample by using the statistical features and the personalized features comprises the following steps:
constructing a feature space of the driving pattern sample by calculating the statistical features of each driving operation parameter in each time window in the driving pattern sample and the personalized features of the driving operation parameters with typical features in each time window; the typical characteristic driving operation parameter includes a steering wheel angle.
2. The driving pattern feature construction and screening method of claim 1, wherein the step of performing a significance analysis on the features in the feature space and reducing the feature dimension of the feature space by eliminating extraneous features comprises:
judging whether the features have significance differences among different driving modes or not by performing significance analysis on the features in the feature space;
and if the characteristic has no significant difference among different driving modes, removing the characteristic from the characteristic space.
3. The driving pattern feature construction and screening method of claim 1, wherein the driving operation parameters in the driving pattern samples include at least an accelerator pedal position, a vehicle speed, a brake pedal position, a steering wheel angle, a lateral acceleration, and a yaw rate.
4. The driving pattern feature construction and screening method of claim 1, wherein the feature screening algorithm further comprises a sequence forward floating selection algorithm.
5. The driving pattern feature construction and screening method of claim 1, wherein a feature dimension of the current optimal feature space is less than or equal to a preset value.
6. A driving pattern feature construction and screening apparatus comprising a processor, a memory and a feature construction and screening program stored on and executable on the memory, the feature construction and screening program when executed by the processor implementing the steps of the driving pattern feature construction and screening method of any one of claims 1 to 5.
7. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a feature construction and screening program, which when executed by a processor, implements the steps of the driving pattern feature construction and screening method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011471000.1A CN112232525B (en) | 2020-12-15 | 2020-12-15 | Driving mode characteristic construction and screening method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011471000.1A CN112232525B (en) | 2020-12-15 | 2020-12-15 | Driving mode characteristic construction and screening method and device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112232525A CN112232525A (en) | 2021-01-15 |
| CN112232525B true CN112232525B (en) | 2021-07-13 |
Family
ID=74123621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011471000.1A Active CN112232525B (en) | 2020-12-15 | 2020-12-15 | Driving mode characteristic construction and screening method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112232525B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095675A (en) * | 2015-09-07 | 2015-11-25 | 浙江群力电气有限公司 | Switch cabinet fault feature selection method and apparatus |
| CN109460780A (en) * | 2018-10-17 | 2019-03-12 | 深兰科技（上海）有限公司 | Safe driving of vehicle detection method, device and the storage medium of artificial neural network |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9490792B2 (en) * | 2010-02-10 | 2016-11-08 | Freescale Semiconductor, Inc. | Pulse width modulation with effective high duty resolution |
| US8988807B1 (en) * | 2013-09-19 | 2015-03-24 | HGST Netherlands B.V. | Disk drive with different data sector integrated preambles in adjacent data tracks |
| US10317392B2 (en) * | 2016-06-23 | 2019-06-11 | Roche Sequencing Solutions, Inc. | Formation and calibration of nanopore sequencing cells |
| CN106971194B (en) * | 2017-02-16 | 2021-02-12 | 江苏大学 | Driving intention recognition method based on improved HMM and SVM double-layer algorithm |
| CN111204348A (en) * | 2020-01-21 | 2020-05-29 | 腾讯云计算（北京）有限责任公司 | Method and device for adjusting vehicle running parameters, vehicle and storage medium |
| CN111950585A (en) * | 2020-06-29 | 2020-11-17 | 广东技术师范大学 | XGboost-based underground comprehensive pipe gallery safety condition assessment method |
| CN111882923A (en) * | 2020-07-15 | 2020-11-03 | 山东省网联智能车辆产业技术研究院有限公司 | Intelligent networking automobile behavior identification method |
-
2020
- 2020-12-15 CN CN202011471000.1A patent/CN112232525B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095675A (en) * | 2015-09-07 | 2015-11-25 | 浙江群力电气有限公司 | Switch cabinet fault feature selection method and apparatus |
| CN109460780A (en) * | 2018-10-17 | 2019-03-12 | 深兰科技（上海）有限公司 | Safe driving of vehicle detection method, device and the storage medium of artificial neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112232525A (en) | 2021-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6966654B2 (en) | Virtual vehicle operation methods, model training methods, operation devices, and storage media | |
| US10677686B2 (en) | Method and apparatus for autonomous system performance and grading | |
| CN110415521B (en) | Traffic data prediction method, apparatus and computer-readable storage medium | |
| CN110569783A (en) | Method and system for identifying lane changing intention of driver | |
| CN107301289B (en) | Method for realizing traffic flow cellular automaton model based on intelligent game | |
| CN110077416B (en) | Decision tree-based driver intention analysis method and system | |
| CN112466118A (en) | Vehicle driving behavior recognition method, system, electronic device and storage medium | |
| WO2023071700A1 (en) | Automatic driving-based data processing method and apparatus | |
| CN112232525B (en) | Driving mode characteristic construction and screening method and device and storage medium | |
| KR20150031051A (en) | Apparatus for judging driver inattention and method thereof | |
| CN113096414B (en) | Intersection timing method, system and device based on traffic conflict analysis | |
| CN112124302B (en) | Automatic parking control method and device, vehicle and storage medium | |
| CN108460057B (en) | User travel mining method and device based on unsupervised learning | |
| CN115366891A (en) | Driving style recognition method, system and storage medium | |
| CN111753926A (en) | Data sharing method and system for smart city | |
| CN112347896A (en) | Head data processing method and device based on multitask neural network | |
| CN112230565A (en) | Method and device for simulating driving of vehicle, electronic equipment and computer-readable storage medium | |
| CN114842205B (en) | Vehicle loss detection method, device, equipment and storage medium | |
| CN110796024B (en) | Automatic driving visual perception test method and device for failure sample | |
| Weber et al. | Runtime optimization of a CNN model for environment perception | |
| CN116776204B (en) | Driver risk sensitivity differentiation characterization method, device, equipment and medium | |
| Zheng et al. | Non-Uniform time window processing of in-vehicle signals for maneuvers recognition and route recovery | |
| CN114550147B (en) | Automobile data acquisition, analysis and processing method and system based on Internet of things | |
| US20230196194A1 (en) | Computer program product and artificial intelligence training control device | |
| CN115292923A (en) | Automatic parking trajectory evaluation method and system based on scene library |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |