CN113635906B - Driving style identification method and device based on local time sequence extraction algorithm - Google Patents
Driving style identification method and device based on local time sequence extraction algorithm Download PDFInfo
- Publication number
- CN113635906B CN113635906B CN202111005909.2A CN202111005909A CN113635906B CN 113635906 B CN113635906 B CN 113635906B CN 202111005909 A CN202111005909 A CN 202111005909A CN 113635906 B CN113635906 B CN 113635906B
- Authority
- CN
- China
- Prior art keywords
- driving
- driving data
- driving style
- initial
- subsequences
- 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
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
- B60W40/09—Driving style or behaviour
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2540/00—Input parameters relating to occupants
- B60W2540/30—Driving style
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a driving style identification method and device based on a local time sequence extraction algorithm, wherein the method comprises the following steps: determining at least two driving style sample sets with different driving styles; dividing each driving style sample set in the at least two driving style sample sets into a plurality of initial time subsequences by adopting a preset time sequence dividing method; extracting a plurality of target time subsequences from the plurality of initial time subsequences by adopting a preset local time sequence extraction algorithm; and constructing a similarity recognition model according to the target time subsequences, and recognizing the driving style of the driving data to be recognized according to the similarity recognition model. The method and the device improve timeliness and robustness of the driving style identification method.
Description
Technical Field
The invention relates to the technical field of autonomous driving, in particular to a driving style identification method and device based on a local time sequence extraction algorithm.
Background
The driving style is a representation of personalized driving of a driver, and research of the driving style comprises a plurality of aspects of concentration degree of the driver, subjective requirement of the driver on the motion state of the vehicle and the like. Because the driving style of a person involves complexity and uncertainty of the person, the related content involved in learning the driving style is relatively wide, including the reaction mechanism of the driver to the traffic environment, the age, the mind, the driving experience, and the like of the driver. As a new evaluation index, the driving style can enable the driving behavior to be integrally interpreted, the early warning or forced execution action sent by the vehicle can meet the wish of the driver, and the acceptance and utilization rate of the driver to the automobile system are improved.
Existing research techniques mainly include machine learning based methods. However, the machine learning-based approach suffers from the following drawbacks: (1) The driving style is required to be identified by constructing a neural network based on the machine learning method, and the neural network has a complex structure, so that the identification timeliness of the driving style is reduced; (2) The machine learning-based method requires a large amount of driving data training samples, so that the recognition process of the driving style is very time-consuming, and the effectiveness is further reduced; (3) Because of the high dimensionality of the driving data and the complex relationship between the variables, the existing machine learning method cannot effectively process the multi-variable driving data, resulting in poor robustness of the machine learning method.
Disclosure of Invention
In view of the foregoing, it is necessary to provide a driving style recognition method and device based on a local time sequence extraction algorithm, so as to solve the technical problems of poor timeliness and poor robustness of the driving style recognition method in the prior art.
In order to solve the technical problems, the invention provides a driving style identification method based on a local time sequence extraction algorithm, which comprises the following steps:
determining at least two driving style sample sets with different driving styles;
dividing each driving style sample set in the at least two driving style sample sets into a plurality of initial time subsequences by adopting a preset time sequence dividing method;
extracting a plurality of target time subsequences from the plurality of initial time subsequences by adopting a preset local time sequence extraction algorithm;
and constructing a similarity recognition model according to the target time subsequences, and recognizing the driving style of the driving data to be recognized according to the similarity recognition model.
In one possible implementation, the preset local time sequence extraction algorithm is a shapelets extraction algorithm.
In one possible implementation, the preset time sequence dividing method is a sliding window method.
In one possible implementation, the determining at least two driving style sample sets with different driving styles includes:
acquiring an initial driving data set;
performing dimension reduction on the initial driving data set by adopting a preset dimension reduction algorithm to generate a driving data set to be processed;
and clustering the data to be processed by adopting a clustering algorithm to generate the at least two driving style sample sets.
In one possible implementation, the initial driving data set includes multidimensional initial driving data of a plurality of drivers; the preset dimension reduction algorithm is a principal component analysis method; the step of adopting a preset dimension reduction algorithm to reduce the dimension of the initial driving data set, and the step of generating the driving data set to be processed comprises the following steps:
constructing an initial driving data matrix according to the initial driving data set, wherein the number of lines of the initial driving data matrix is equal to the dimension of the multidimensional initial driving data, and the number of columns of the initial driving data matrix is equal to the number of the plurality of drivers;
zero-equalizing each row of the initial driving data matrix to generate a zero-mean matrix;
calculating a covariance matrix of the zero-mean matrix;
calculating a plurality of eigenvalues of the covariance matrix and a plurality of eigenvectors corresponding to the eigenvalues one by one;
arranging the plurality of eigenvectors into an alternative matrix according to the sequence of the eigenvalues from large to small;
and selecting a first threshold row from the alternative matrix to generate the driving data set to be processed.
In one possible implementation, after the generating the at least two driving style sample sets, the method further includes:
acquiring a driving data verification set;
and verifying the accuracy of the at least two driving style sample sets through the driving data verification set.
In one possible implementation manner, the constructing a similarity recognition model according to the plurality of target time subsequences includes:
symbolizing the plurality of target time subsequences by adopting a symbol set approximation algorithm to generate a plurality of target character strings;
and calculating a plurality of TF-IDF weight vectors of the character strings, and generating the similarity recognition model according to the TF-IDF weight vectors.
In one possible implementation manner, the identifying the driving style of the driving data to be identified according to the similarity identification model includes:
symbolizing the driving data to be identified by adopting a symbol set approximation algorithm to generate a plurality of driving character strings;
calculating the frequency of each driving character string in the plurality of driving character strings to generate a frequency vector;
calculating a plurality of cosine similarity values of the frequency vector and the TF-IDF weight vectors;
and determining the maximum cosine similarity value in the cosine similarity values, determining the most similar target time subsequence corresponding to the maximum cosine similarity value, and determining the driving style of the driving style sample set corresponding to the most similar target time subsequence as the driving style of the driving data to be identified.
In one possible implementation, the symbolizing the plurality of target time sub-sequences using a symbol set approximation algorithm, generating a plurality of target character strings includes:
normalizing the plurality of target time sub-sequences to generate a plurality of normalized target time sub-sequences; the mean value of the plurality of standardized target subsequences is 0, and the standard deviation is 1;
performing dimension reduction processing on a plurality of standardized target subsequences based on a piecewise cumulative approximation method to generate a plurality of dimension reduction subsequences;
and representing the plurality of dimension-reducing subsequences by characters, and generating the plurality of character strings.
On the other hand, the invention also provides a driving style recognition device based on the local time sequence extraction algorithm, which comprises the following steps:
the sample set determining unit is used for determining at least two driving style sample sets with different driving styles;
a sample set dividing unit, configured to divide each driving style sample set in the at least two driving style sample sets into a plurality of initial time sub-sequences by using a preset time sequence dividing method;
a local time sequence extraction unit, configured to extract a plurality of target time subsequences from the plurality of initial time subsequences by using a preset local time sequence extraction algorithm;
and the driving style recognition unit is used for constructing a similarity recognition model according to the plurality of target time subsequences, recognizing driving data to be recognized according to the similarity recognition model, and recognizing the driving style of the driving data to be recognized.
The beneficial effects of adopting the embodiment are as follows: according to the driving style identification method based on the local time sequence extraction algorithm, the driving style of the driving data to be identified is identified by constructing the similarity identification model without building a neural network model, so that the identification model is light, the speed of constructing the similarity identification model is improved, and therefore the timeliness of identifying the driving style of the driving data to be identified is improved. Further, by sequentially adopting a preset time sequence segmentation method and a preset local time sequence extraction algorithm to extract a plurality of target time subsequences from at least two driving style sample sets, dimension reduction can be realized on driving style sample data in the driving style sample sets, so that the speed of constructing a similarity recognition model is further improved, and the timeliness of driving style recognition of driving data to be recognized is further improved. Furthermore, the driving style sample data in the driving style sample set is subjected to dimension reduction by adopting a preset time sequence segmentation method and a preset local time sequence extraction algorithm in sequence, so that redundant sequences and noise in the driving style sample data are reduced, complex relations between high dimension and variables can be processed, and the robustness of the driving style identification method is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an embodiment of a driving style recognition method based on a local time sequence extraction algorithm provided by the invention;
FIG. 2 is a flow chart of the embodiment of S101 in FIG. 1 according to the present invention;
FIG. 3 is a flow chart of the embodiment of S202 in FIG. 2 according to the present invention;
FIG. 4 is a flowchart illustrating an embodiment of the present invention after step S203;
FIG. 5 is a flowchart illustrating an embodiment of the method for constructing the similarity recognition model in S104 of FIG. 1 according to the present invention;
FIG. 6 is a flowchart illustrating the process of S501 in FIG. 5 according to an embodiment of the present invention;
FIG. 7 is a schematic structural diagram of an embodiment of the step S601 of the present invention;
FIG. 8 is a flowchart illustrating an embodiment of identifying driving style of the driving data to be identified according to the similarity identification model in S104 of FIG. 1 according to the present invention;
FIG. 9 is a schematic structural diagram of an embodiment of a driving style recognition device based on a local time sequence extraction algorithm according to the present invention;
fig. 10 is a schematic structural diagram of an embodiment of an electronic device according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the embodiments of the present application, unless otherwise indicated, the meaning of "a plurality" is two or more. "and/or", describes an association relationship of an associated object, meaning that there may be three relationships, for example: a and/or B may represent: a exists alone, A and B exist together, and B exists alone.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
The invention provides a driving style recognition method and device based on a local time sequence extraction algorithm, and the driving style recognition method and device are respectively described below.
Fig. 1 is a schematic flow chart of an embodiment of a driving style recognition method based on a local time sequence extraction algorithm provided by the present invention, and as shown in fig. 1, the driving style recognition method based on the local time sequence extraction algorithm includes:
s101, determining at least two driving style sample sets with different driving styles;
s102, dividing each driving style sample set in at least two driving style sample sets into a plurality of initial time subsequences by adopting a preset time sequence dividing method;
s103, extracting a plurality of target time subsequences from a plurality of initial time subsequences by adopting a preset local time sequence extraction algorithm;
s104, constructing a similarity recognition model according to the plurality of target time subsequences, and recognizing the driving style of the driving data to be recognized according to the similarity recognition model.
Compared with the prior art, the driving style identification method based on the local time sequence extraction algorithm provided by the embodiment of the invention identifies the driving style of the driving data to be identified by constructing the similarity identification model without building a neural network model, so that the identification model is light, the speed of constructing the similarity identification model is improved, and the timeliness of identifying the driving style of the driving data to be identified is improved. Further, by sequentially adopting a preset time sequence segmentation method and a preset local time sequence extraction algorithm to extract a plurality of target time subsequences from at least two driving style sample sets, dimension reduction can be realized on driving style sample data in the driving style sample sets, so that the speed of constructing a similarity recognition model is further improved, and the timeliness of driving style recognition of driving data to be recognized is further improved. Furthermore, the driving style sample data in the driving style sample set is subjected to dimension reduction by adopting a preset time sequence segmentation method and a preset local time sequence extraction algorithm in sequence, so that redundant sequences and noise in the driving style sample data are reduced, complex relations between high dimension and variables can be processed, and the robustness of the driving style identification method is improved.
In some embodiments of the present invention, the driving style may include a conservative type and an aggressive type, and the at least two driving style sample sets in step S101 may include a conservative type sample set and an aggressive type sample set.
In some other embodiments of the present invention, the driving style may include a conservative type, an intermediate type, and an aggressive type, and the at least two driving style sample sets in step S101 may include a conservative type sample set, an intermediate type sample set, and an aggressive type sample set.
It should be understood that: the driving style may include more types, and may be adjusted according to actual situations, which will not be described herein.
In some embodiments of the present invention, the time series segmentation method preset in step S102 is a sliding window method, specifically, the division of the driving style sample set is achieved by sliding a sliding window along a time direction, where parameters of the sliding window include a window length and a sliding step length, the window length is used to represent a data extraction range of the sliding window, and the sliding step length is used to represent a length of each sliding of the sliding window.
Through the sliding window method in the embodiment, each driving style sample set style can be quickly divided into a plurality of initial time subsequences, and the recognition speed of the driving style is further improved.
In one embodiment of the present invention, the sliding window has a window length of 60 seconds and a sliding step length of 30 seconds, namely: the two adjacent sliding windows have overlapping areas, so that data leakage can be avoided, and the reliability of a sliding window method is improved.
In some embodiments of the present invention, the local time series extraction algorithm preset in step S203 is a shape extraction algorithm.
Because the shapelets extraction algorithm has the characteristic of strong interpretability, the interpretability of the recognition result can be improved compared with the prior art by setting the preset local time sequence extraction algorithm as the shapelets extraction algorithm. Namely: it is possible to explain why the driving data to be recognized is classified into a specific driving style, so that the driver can understand and receive it easily.
In some embodiments of the present invention, as shown in fig. 2, step S101 includes:
s201, acquiring an initial driving data set;
s202, adopting a preset dimension reduction algorithm to reduce the dimension of the initial driving data set, and generating a driving data set to be processed;
and S203, clustering the data to be processed by adopting a clustering algorithm to generate at least two driving style sample sets.
According to the embodiment of the invention, the dimension of the driving style sample data in the driving style sample set can be reduced by adopting the preset dimension reduction algorithm to reduce the dimension of the initial driving data set, and the speed of obtaining at least two driving style sample sets is further improved, so that the timeliness of driving style identification of the driving data to be identified can be further improved. Furthermore, the driving style sample data in the driving style sample set is subjected to dimension reduction, and redundant sequences and noise in the driving style sample data can be further reduced, so that the driving data set with high dimension and complex relationship between variables can be processed, and the robustness of the driving style identification method is further improved.
In some embodiments of the invention, the initial driving data set in step S201 may be obtained by a plurality of sensors, for example: steering wheel sensors, brake sensors, throttle sensors, etc.
And the initial driving data set may include data generated by the vehicle at the time of actual use or at the time of testing.
It should be understood that the clustering algorithm in step S203 includes a number of cluster centroids equal to at least two driving style sample sets. Namely: each cluster centroid corresponds to a driving style sample set.
In some embodiments of the present invention, the dimension reduction algorithm preset in step S202 may be any one of principal component analysis, linear discriminant analysis, local linear embedding, and laplace feature mapping.
In a preferred embodiment of the present invention, in order to increase the dimension reduction speed, a preset dimension reduction algorithm is a principal component analysis method.
In a specific embodiment of the invention, the initial driving data set comprises multidimensional initial driving data of a plurality of drivers; as shown in fig. 3, step S202 includes:
s301, constructing an initial driving data matrix according to an initial driving data set; the number of lines of the initial driving data matrix is equal to the number of dimensions of the multidimensional initial driving data, and the number of columns of the initial driving data matrix is equal to the number of a plurality of drivers;
s302, carrying out zero-mean on each row of the initial driving data matrix to generate a zero-mean matrix;
s303, calculating a covariance matrix of the zero-mean matrix;
s304, calculating a plurality of eigenvalues of the covariance matrix and a plurality of eigenvectors corresponding to the eigenvalues one by one;
s305, arranging a plurality of eigenvectors into an alternative matrix according to the sequence of the eigenvalues from large to small;
s306, selecting a first threshold line from the alternative matrix, and generating a driving data set to be processed.
In some embodiments of the present invention, the specific process of zero-averaging in step S302 is: the average value of each row of the initial driving data matrix is subtracted from each initial driving data in that row.
It should be understood that: the first threshold is less than the dimension of the multi-dimensional initial driving data.
In order to ensure the reliability of at least two driving style sample sets, in some embodiments of the present invention, as shown in fig. 4, after step S203, further includes:
s401, acquiring a driving data verification set;
s402, verifying the accuracy of at least two driving style sample sets through a driving data verification set.
It should be understood that: step S102 is performed when the accuracy of at least two driving style sample sets is verified to be greater than the threshold accuracy by the driving data verification set, thereby ensuring the recognition accuracy of the constructed similarity recognition model.
In some embodiments of the present invention, the driving data verification set of step S401 may be obtained by means of a questionnaire.
In some embodiments of the present invention, as shown in fig. 5, constructing a similarity recognition model according to the plurality of target time subsequences in step S104 includes:
s501, symbolizing a plurality of target time subsequences by adopting a symbol set approximation (Symbolic Aggregate Approximation, SAX) algorithm to generate a plurality of target character strings;
s502, calculating a plurality of TF-IDF weight vectors of a plurality of character strings, and generating a similarity recognition model according to the plurality of TF-IDF weight vectors.
According to the embodiment of the invention, the SAX algorithm is adopted to symbolize a plurality of target time subsequences to generate a plurality of target character strings, so that the dimension of the plurality of target time subsequences can be reduced, and the timeliness of identifying the driving style of the driving data to be identified is further improved. Furthermore, the dimension reduction is realized on the plurality of target time subsequences, and redundant sequences and noise in the plurality of target subsequences can be reduced, so that the embodiment of the invention can process a driving data set with high dimension and complex relationship between variables, and the robustness of the driving style recognition method based on the local time sequence extraction algorithm is further improved.
In some embodiments of the present invention, as shown in fig. 6, step S501 includes:
s601, normalizing a plurality of target time sub-sequences to generate a plurality of normalized target time sub-sequences; the mean value of the plurality of standardized target subsequences is 0, and the standard deviation is 1;
s602, performing dimension reduction processing on a plurality of standardized target subsequences based on a piecewise accumulation approximation method to generate a plurality of dimension reduction subsequences;
s603, representing the plurality of dimension-reducing subsequences by characters, and generating a plurality of character strings.
In a specific embodiment of the present invention, as shown in fig. 7, the abscissa X represents a sample point of the target time sub-sequence data, the ordinate Y represents a value corresponding to the normalized sample point of the target time sub-sequence, in fig. 7, the target time sub-sequence is equally probability-divided by using 4 horizontal dashed lines, 5 intervals are total, and each interval is respectively assigned with a corresponding character A, B, C, D, E, the solid curve in the figure is one normalized target sub-sequence of the normalized target sub-sequences, the short horizontal line is a plurality of dimension-reduced sub-sequences after dimension-reduction processing, and after steps S601-S603, a character string corresponding to one of the target time sub-sequences is "CDEEEDCB".
In some embodiments of the present invention, as shown in fig. 8, identifying the driving style of the driving data to be identified according to the similarity identification model in step S104 includes:
s801, symbolizing driving data to be identified by adopting a symbol set approximation algorithm to generate a plurality of driving character strings;
s802, calculating the frequency of each driving character string in the plurality of driving character strings, and generating a frequency vector;
s803, calculating a plurality of cosine similarity values of the frequency vector and a plurality of TF-IDF weight vectors;
s804, determining the maximum cosine similarity value in the cosine similarity values, determining the most similar target time subsequence corresponding to the maximum cosine similarity value, and determining the driving style of the driving style sample set corresponding to the most similar target time subsequence as the driving style of the driving data to be identified.
It should be understood that: the process of step S801 is the same as that of step S501, and will not be described here.
In order to better implement the driving style recognition method based on the local time sequence extraction algorithm in the embodiment of the present invention, correspondingly, as shown in fig. 9, the embodiment of the present invention further provides a driving style recognition device 900 based on the local time sequence extraction algorithm, which includes:
a sample set determining unit 901 for determining at least two driving style sample sets having different driving styles;
a sample set dividing unit 902, configured to divide each of at least two driving style sample sets into a plurality of initial time sub-sequences by using a preset time sequence dividing method;
a local time sequence extracting unit 903, configured to extract a plurality of target time sub-sequences from a plurality of initial time sub-sequences by using a preset local time sequence extracting algorithm;
the driving style recognition unit 904 is configured to construct a similarity recognition model according to the multiple target time subsequences, and recognize driving data to be recognized according to the similarity recognition model, so as to recognize the driving style of the driving data to be recognized.
The driving style recognition device 900 based on the local time sequence extraction algorithm provided in the foregoing embodiment may implement the technical solution described in the foregoing embodiment of the driving style recognition method based on the local time sequence extraction algorithm, and the specific implementation principle of each module or unit may refer to the corresponding content in the foregoing embodiment of the driving style recognition method based on the local time sequence extraction algorithm, which is not described herein again.
As shown in fig. 10, the present invention further provides an electronic device 1000 accordingly. The electronic device 1000 comprises a processor 1001, a memory 1002 and a display 1003. Fig. 10 shows only some of the components of the electronic device 1000, but it should be understood that not all of the illustrated components are required to be implemented and that more or fewer components may be implemented instead.
The memory 1002 may be an internal storage unit of the electronic device 1000 in some embodiments, such as a hard disk or memory of the electronic device 1000. The memory 1002 may also be an external storage device of the electronic device 1000 in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the electronic device 1000.
Further, the memory 1002 may also include both internal storage units and external storage devices of the electronic device 1000. The memory 1002 is used for storing application software and various types of data for installing the electronic device 1000.
The processor 1001 may in some embodiments be a central processing unit (Central Processing Unit, CPU), microprocessor or other data processing chip for executing program code or processing data stored in the memory 1002, such as the driving style recognition method based on a local time series extraction algorithm in the present invention.
The display 1003 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, or the like in some embodiments. The display 1003 is used for displaying information at the electronic device 1000 and for displaying a visualized user interface. The components 1001-1003 of the electronic device 1000 communicate with each other over a system bus.
In an embodiment, when the processor 1001 executes the driving style recognition program based on the local time series extraction algorithm in the memory 1002, the following steps may be implemented:
determining at least two driving style sample sets with different driving styles;
dividing each of at least two driving style sample sets into a plurality of initial time subsequences by adopting a preset time sequence dividing method;
extracting a plurality of target time subsequences from a plurality of initial time subsequences by adopting a preset local time sequence extraction algorithm;
and constructing a similarity recognition model according to the multiple target time subsequences, and recognizing the driving style of the driving data to be recognized according to the similarity recognition model.
It should be understood that: the processor 1001 may, in executing the driving style recognition program based on the local time series extraction algorithm in the memory 1002, realize other functions in addition to the above functions, and in particular, see the description of the corresponding method embodiments above.
Further, the type of the electronic device 1000 is not particularly limited, and the electronic device 1000 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (personal digital assistant, PDA), a wearable device, a laptop (laptop), etc. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices that carry IOS, android, microsoft or other operating systems. The portable electronic device described above may also be other portable electronic devices, such as a laptop computer (laptop) or the like having a touch-sensitive surface, e.g. a touch panel. It should also be appreciated that in other embodiments of the invention, the electronic device 1000 may not be a portable electronic device, but rather a desktop computer having a touch-sensitive surface (e.g., a touch panel).
Accordingly, the embodiments of the present application further provide a computer readable storage medium, where the computer readable storage medium is used to store a computer readable program or instructions, and when the program or instructions are executed by a processor, the method steps or functions provided in the foregoing method embodiments can be implemented.
It will be appreciated by those skilled in the art that the whole or part of the flow of the method of the above embodiment may be implemented by a computer program to instruct related hardware, and a driving style recognition program based on a local time series extraction algorithm may be stored in a computer readable storage medium. The computer readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory.
The driving style recognition method and device based on the local time sequence extraction algorithm provided by the invention are described in detail, and specific examples are applied to the description of the principle and the implementation mode of the invention, and the description of the examples is only used for helping to understand the method and the core idea of the invention; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in light of the ideas of the present invention, the present description should not be construed as limiting the present invention.
Claims (6)
1. A driving style recognition method based on a local time sequence extraction algorithm, comprising:
determining at least two driving style sample sets with different driving styles;
dividing each driving style sample set in the at least two driving style sample sets into a plurality of initial time subsequences by adopting a preset time sequence dividing method;
extracting a plurality of target time subsequences from the plurality of initial time subsequences by adopting a preset local time sequence extraction algorithm;
constructing a similarity recognition model according to the target time subsequences, and recognizing the driving style of the driving data to be recognized according to the similarity recognition model;
the determining at least two driving style sample sets with different driving styles comprises:
acquiring an initial driving data set;
performing dimension reduction on the initial driving data set by adopting a preset dimension reduction algorithm to generate a driving data set to be processed;
clustering the driving data sets to be processed by adopting a clustering algorithm to generate at least two driving style sample sets;
the initial driving data set includes multidimensional initial driving data of a plurality of drivers; the preset dimension reduction algorithm is a principal component analysis method; the step of adopting a preset dimension reduction algorithm to reduce the dimension of the initial driving data set, and the step of generating the driving data set to be processed comprises the following steps:
constructing an initial driving data matrix according to the initial driving data set, wherein the number of lines of the initial driving data matrix is equal to the dimension of the multidimensional initial driving data, and the number of columns of the initial driving data matrix is equal to the number of the plurality of drivers;
zero-equalizing each row of the initial driving data matrix to generate a zero-mean matrix;
calculating a covariance matrix of the zero-mean matrix;
calculating a plurality of eigenvalues of the covariance matrix and a plurality of eigenvectors corresponding to the eigenvalues one by one;
arranging the plurality of eigenvectors into an alternative matrix according to the sequence of the eigenvalues from large to small;
selecting a first threshold row from the alternative matrix to generate the driving data set to be processed;
the constructing a similarity recognition model according to the plurality of target time subsequences includes:
symbolizing the plurality of target time subsequences by adopting a symbol set approximation algorithm to generate a plurality of target character strings;
calculating a plurality of TF-IDF weight vectors of the target character strings, and generating the similarity recognition model according to the TF-IDF weight vectors;
the identifying the driving style of the driving data to be identified according to the similarity identification model comprises the following steps:
symbolizing the driving data to be identified by adopting a symbol set approximation algorithm to generate a plurality of driving character strings;
calculating the frequency of each driving character string in the plurality of driving character strings to generate a frequency vector;
calculating a plurality of cosine similarity values of the frequency vector and the TF-IDF weight vectors;
and determining the maximum cosine similarity value in the cosine similarity values, determining the most similar target time subsequence corresponding to the maximum cosine similarity value, and determining the driving style of the driving style sample set corresponding to the most similar target time subsequence as the driving style of the driving data to be identified.
2. The driving style recognition method based on the local time series extraction algorithm according to claim 1, wherein the preset local time series extraction algorithm is a shape extraction algorithm.
3. The driving style recognition method based on the local time series extraction algorithm according to claim 1, wherein the preset time series segmentation method is a sliding window method.
4. The driving style identification method based on a local time series extraction algorithm according to claim 1, further comprising, after said generating said at least two driving style sample sets:
acquiring a driving data verification set;
and verifying the accuracy of the at least two driving style sample sets through the driving data verification set.
5. The driving style recognition method based on the local time series extraction algorithm according to claim 1, wherein the symbolizing the plurality of target time sub-series using a symbol set approximation algorithm, generating a plurality of target character strings, comprises:
normalizing the plurality of target time sub-sequences to generate a plurality of normalized target time sub-sequences; the mean value of the plurality of standardized target subsequences is 0, and the standard deviation is 1;
performing dimension reduction processing on a plurality of standardized target subsequences based on a piecewise cumulative approximation method to generate a plurality of dimension reduction subsequences;
and representing the plurality of dimension-reducing subsequences by characters, and generating the plurality of target character strings.
6. A driving style recognition device based on a local time series extraction algorithm, comprising:
the sample set determining unit is used for determining at least two driving style sample sets with different driving styles;
a sample set dividing unit, configured to divide each driving style sample set in the at least two driving style sample sets into a plurality of initial time sub-sequences by using a preset time sequence dividing method;
a local time sequence extraction unit, configured to extract a plurality of target time subsequences from the plurality of initial time subsequences by using a preset local time sequence extraction algorithm;
the driving style recognition unit is used for constructing a similarity recognition model according to the plurality of target time subsequences, recognizing driving data to be recognized according to the similarity recognition model, and recognizing the driving style of the driving data to be recognized;
the determining at least two driving style sample sets with different driving styles comprises:
acquiring an initial driving data set;
performing dimension reduction on the initial driving data set by adopting a preset dimension reduction algorithm to generate a driving data set to be processed;
clustering the driving data sets to be processed by adopting a clustering algorithm to generate at least two driving style sample sets;
the initial driving data set includes multidimensional initial driving data of a plurality of drivers; the preset dimension reduction algorithm is a principal component analysis method; the step of adopting a preset dimension reduction algorithm to reduce the dimension of the initial driving data set, and the step of generating the driving data set to be processed comprises the following steps:
constructing an initial driving data matrix according to the initial driving data set, wherein the number of lines of the initial driving data matrix is equal to the dimension of the multidimensional initial driving data, and the number of columns of the initial driving data matrix is equal to the number of the plurality of drivers;
zero-equalizing each row of the initial driving data matrix to generate a zero-mean matrix;
calculating a covariance matrix of the zero-mean matrix;
calculating a plurality of eigenvalues of the covariance matrix and a plurality of eigenvectors corresponding to the eigenvalues one by one;
arranging the plurality of eigenvectors into an alternative matrix according to the sequence of the eigenvalues from large to small;
selecting a first threshold row from the alternative matrix to generate the driving data set to be processed;
the constructing a similarity recognition model according to the plurality of target time subsequences includes:
symbolizing the plurality of target time subsequences by adopting a symbol set approximation algorithm to generate a plurality of target character strings;
calculating a plurality of TF-IDF weight vectors of the target character strings, and generating the similarity recognition model according to the TF-IDF weight vectors;
the identifying the driving style of the driving data to be identified according to the similarity identification model comprises the following steps:
symbolizing the driving data to be identified by adopting a symbol set approximation algorithm to generate a plurality of driving character strings;
calculating the frequency of each driving character string in the plurality of driving character strings to generate a frequency vector;
calculating a plurality of cosine similarity values of the frequency vector and the TF-IDF weight vectors;
and determining the maximum cosine similarity value in the cosine similarity values, determining the most similar target time subsequence corresponding to the maximum cosine similarity value, and determining the driving style of the driving style sample set corresponding to the most similar target time subsequence as the driving style of the driving data to be identified.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111005909.2A CN113635906B (en) | 2021-08-30 | 2021-08-30 | Driving style identification method and device based on local time sequence extraction algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111005909.2A CN113635906B (en) | 2021-08-30 | 2021-08-30 | Driving style identification method and device based on local time sequence extraction algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113635906A CN113635906A (en) | 2021-11-12 |
CN113635906B true CN113635906B (en) | 2023-07-25 |
Family
ID=78424394
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111005909.2A Active CN113635906B (en) | 2021-08-30 | 2021-08-30 | Driving style identification method and device based on local time sequence extraction algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113635906B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
CN103786733A (en) * | 2013-12-27 | 2014-05-14 | 宁波大学 | Environment-friendly driving behavior prompting method for automatic transmission automobile |
KR20200076129A (en) * | 2018-12-19 | 2020-06-29 | 한양대학교 산학협력단 | LSTM-based steering behavior monitoring device and its method |
CN112036297A (en) * | 2020-08-28 | 2020-12-04 | 长安大学 | Typical and extreme scene division and extraction method based on internet vehicle driving data |
DE102019211017A1 (en) * | 2019-07-25 | 2021-01-14 | Zf Friedrichshafen Ag | Method for clustering different time series values of vehicle data and use of the method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107679557B (en) * | 2017-09-19 | 2020-11-27 | 平安科技(深圳)有限公司 | Driving model training method, driver identification method, device, equipment and medium |
-
2021
- 2021-08-30 CN CN202111005909.2A patent/CN113635906B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101633359A (en) * | 2008-07-24 | 2010-01-27 | 通用汽车环球科技运作公司 | Adaptive vehicle control system with driving style recognition |
CN103786733A (en) * | 2013-12-27 | 2014-05-14 | 宁波大学 | Environment-friendly driving behavior prompting method for automatic transmission automobile |
KR20200076129A (en) * | 2018-12-19 | 2020-06-29 | 한양대학교 산학협력단 | LSTM-based steering behavior monitoring device and its method |
DE102019211017A1 (en) * | 2019-07-25 | 2021-01-14 | Zf Friedrichshafen Ag | Method for clustering different time series values of vehicle data and use of the method |
CN112036297A (en) * | 2020-08-28 | 2020-12-04 | 长安大学 | Typical and extreme scene division and extraction method based on internet vehicle driving data |
Also Published As
Publication number | Publication date |
---|---|
CN113635906A (en) | 2021-11-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108564954B (en) | Deep neural network model, electronic device, identity verification method, and storage medium | |
US9911052B2 (en) | System and method for superimposed handwriting recognition technology | |
US9448651B2 (en) | Intelligent touchscreen keyboard with finger differentiation | |
CN110033018B (en) | Graph similarity judging method and device and computer readable storage medium | |
US10007859B2 (en) | System and method for superimposed handwriting recognition technology | |
CN113056743A (en) | Training neural networks for vehicle re-recognition | |
CN103473492A (en) | Method and user terminal for recognizing permission | |
CN112214402B (en) | Code verification algorithm selection method, device and storage medium | |
CN112070506A (en) | Risk user identification method, device, server and storage medium | |
JP6281491B2 (en) | Text mining device, text mining method and program | |
CN111881683A (en) | Method and device for generating relation triples, storage medium and electronic equipment | |
CN110850982B (en) | AR-based man-machine interaction learning method, system, equipment and storage medium | |
CN113635906B (en) | Driving style identification method and device based on local time sequence extraction algorithm | |
CN116775918B (en) | Cross-modal retrieval method, system, equipment and medium based on complementary entropy contrast learning | |
CN110852261B (en) | Target detection method and device, electronic equipment and readable storage medium | |
CN107912062B (en) | System and method for overlaying handwriting | |
CN111078744A (en) | Method, equipment and storage medium for pre-docking and guiding scientific and technological requirements | |
US11854528B2 (en) | Method and system for detecting unsupported utterances in natural language understanding | |
US20180349739A1 (en) | Method and apparatus for recognizing object based on vocabulary tree | |
Singh et al. | A Temporal Convolutional Network for modeling raw 3D sequences and air-writing recognition | |
CN107995181B (en) | Gait-based identity authentication method, device, equipment and storage medium | |
KR102413181B1 (en) | Mobile device and method for determining user-related information using drawing data generated by the user | |
CN114742060B (en) | Entity identification method, entity identification device, electronic equipment and storage medium | |
CN113033512B (en) | Narrow-strip-shaped fingerprint identification method, storage medium and electronic equipment | |
CN109933788B (en) | Type determining method, device, equipment and medium |
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 |