A method, device, equipment and storage medium for generating vehicle driving data

Abstract

The present invention discloses a method, device, equipment, and storage medium for generating vehicle driving data. The method comprises: obtaining raw GPS data of a vehicle on each road section and the road length of the road section; dividing kinematic segments and performing clustering to classify the kinematic segments into multiple operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed and time in combination with the corresponding road speed; classifying the road section into multiple road conditions and calculating the displacement generated by the kinematic segments of each operating condition type; inputting the road length, road condition, and displacement into a generative adversarial network for training, and obtaining a position sequence through the trained generative adversarial network; and performing transition processing on the position sequence to obtain driving data of the vehicle under specific road conditions. The present invention can generate vehicle driving data similar to that of a vehicle on a real road based on road length, road speed, and road condition, effectively improving the authenticity and accuracy of the generated data.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile data processing, and in particular to a method, device, equipment and storage medium for generating automobile driving data. Background Art

[0002] Current technologies for generating automotive driving data primarily focus on generating driving cycle data. Driving cycles refer to the operating conditions of a vehicle during transportation and transportation. These include starting, acceleration, constant speed, deceleration, cornering, uphill and downhill driving, and parking. A driving cycle, also known as a vehicle test cycle, describes the speed-time curve of a vehicle's travel and reflects its kinematic characteristics.

[0003] Existing vehicle driving cycle data, primarily used to test various aspects of vehicle performance, is a critical, common foundational technology for the automotive industry. It forms the basis for vehicle energy consumption and emission testing methods and limit standards, and serves as the primary benchmark for calibrating and optimizing various vehicle performance indicators. The test results are also available to consumers and published in the market. However, the performance data derived from these driving cycle tests deviates somewhat from the vehicle's actual performance. Because these tests are conducted only on typical driving situations, the data generated is primarily a simple splicing of multiple typical motion clips, with details generated based on a single condition. However, the use of a single feature to describe the driving scene results in a single-feature focus during data processing, classification, and learning. Consequently, the resulting driving cycle data differs significantly from actual road driving data. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a method, device, equipment and storage medium for generating automobile driving data, which can generate automobile driving data similar to the driving conditions of a car on a real road based on road length, road speed and road conditions, thereby effectively improving the authenticity and accuracy of the generated data.

[0005] To achieve the above objectives, an embodiment of the present invention provides a method for generating vehicle driving data, comprising:

[0006] Obtain the original GPS data of the car traveling on each road segment and the road length of each road segment;

[0007] Dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments;

[0008] performing a cluster analysis on the plurality of kinematic segments, classifying the plurality of kinematic segments into a plurality of operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time in combination with a road speed corresponding to each operating condition type;

[0009] According to the proportion of kinematic segments of various working conditions in each road segment data, the road segment is divided into multiple road conditions, and the displacement generated by the kinematic segments of various working conditions in each road segment data is calculated;

[0010] Training a generative adversarial network constructed using the road length, the road condition, and the displacement inputs, and obtaining position sequences of various working condition types through the trained generative adversarial network;

[0011] The kinematic segments in the position sequence are subjected to transition processing to obtain driving data of the vehicle under specific road conditions.

[0012] As an improvement to the above solution, the division of the original GPS data into kinematic segments is specifically as follows:

[0013] All extreme points in the original GPS data are extracted, and kinematic segments are divided according to the extreme points.

[0014] As an improvement to the above solution, after dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments, the method further includes:

[0015] calculating the acceleration of each kinematic segment and determining whether there is abnormal acceleration;

[0016] If abnormal acceleration exists, the kinematic segment corresponding to the abnormal acceleration is removed.

[0017] As an improvement to the above solution, cluster analysis is performed on the multiple kinematic segments to classify the multiple kinematic segments into multiple operating condition types, and the kinematic segments of each operating condition type are fitted into a function of road speed with respect to time in combination with the road speed corresponding to each operating condition type, specifically including:

[0018] Performing cluster analysis on the multiple kinematic segments using a K-means algorithm to classify the multiple kinematic segments into multiple operating condition types; wherein the operating condition types include at least constant speed driving, overtaking, avoiding, and traffic lights;

[0019] The classification center point of each operating condition type and the N points closest to the classification center point are selected, and the kinematic segment of each operating condition type is fitted into a function of road speed with respect to time in combination with the road speed corresponding to each operating condition type; where N is a positive integer.

[0020] As an improvement to the above solution, the road section is divided into multiple road conditions according to the proportion of kinematic segments of various working conditions in each road section data, and the displacement generated by the kinematic segments of various working conditions in each road section data is calculated, specifically including:

[0021] Calculate the proportion of kinematic segments of various working condition types in each road segment data;

[0022] Using a K-means algorithm to classify the road section into multiple road conditions according to the ratio; wherein the road conditions include at least smooth, slow, congested, and severely congested;

[0023] Calculate the displacements generated by the kinematic segments of various operating conditions in each road segment data.

[0024] As an improvement to the above solution, the training method of the generative adversarial network specifically includes:

[0025] Inputting the road length, the road condition, and the displacement into a generator in a constructed generative adversarial network, the generator generating an initial position sequence of various working condition types;

[0026] Inputting the initial position sequence into a discriminator in the generative adversarial network, and allowing the discriminator to determine whether the initial position sequence conforms to actual driving conditions;

[0027] The generator and the discriminator are iteratively updated according to the judgment result until the discriminator determines that the initial position sequence generated by the generator conforms to the actual driving situation, thereby obtaining a trained generative adversarial network.

[0028] As an improvement to the above solution, the kinematic segments in the position sequence are subjected to transition processing to obtain driving data of the vehicle under specific road conditions, specifically including:

[0029] Extracting two node data before and after the connection point of any two adjacent kinematic segments in the position sequence;

[0030] Newton interpolation is used to supplement the intermediate data between the two node data to achieve a smooth transition between any two adjacent kinematic segments, thereby obtaining the driving data of the car under specific road conditions.

[0031] An embodiment of the present invention further provides a device for generating vehicle driving data, comprising:

[0032] An acquisition module is used to obtain the original GPS data of the car traveling on each road section and the road length of each road section;

[0033] A kinematic segment division module is used to divide the original GPS data into kinematic segments to obtain a plurality of kinematic segments;

[0034] a first clustering module, configured to perform cluster analysis on the plurality of kinematic segments, classify the plurality of kinematic segments into a plurality of operating condition types, and fit the kinematic segments of each operating condition type into a function of road speed with respect to time based on a road speed corresponding to each operating condition type;

[0035] The second clustering module is used to classify the road section into multiple road conditions according to the proportion of kinematic segments of various working condition types in the data of each road section, and calculate the displacement generated by the kinematic segments of various working condition types in the data of each road section;

[0036] a training module, configured to train a generative adversarial network constructed by inputting the road length, the road condition, and the displacement, and obtain position sequences of various working condition types through the trained generative adversarial network;

[0037] The transition processing module is used to perform transition processing on the kinematic segments in the position sequence to obtain driving data of the vehicle under specific road conditions.

[0038] An embodiment of the present invention also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for generating vehicle driving data described in any one of the above items is implemented.

[0039] An embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any of the above-mentioned methods for generating automobile driving data.

[0040] Compared to the prior art, the beneficial effects of the method, device, equipment, and storage medium for generating automobile driving data provided by the embodiments of the present invention are as follows: by selecting real driving data of automobiles on different road sections and road length characteristics to train the generative adversarial network, the characteristics of its adversarial training are fully utilized to achieve better driving data generation effects. During the data processing stage, the road condition characteristics in each data are extracted, and the road length, road speed, and road conditions are combined to generate automobile driving data, which is closer to the actual driving situation and effectively improves the authenticity and accuracy of the generated data. In addition, the generative adversarial network can flexibly adjust the input road length, road speed, and road conditions according to different data requirements to generate automobile driving data under specific road conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of a preferred embodiment of a method for generating vehicle driving data provided by the present invention;

[0042] Figure 2 This is an example diagram of the working condition type of a kinematic segment in a method for generating automobile driving data provided by the present invention;

[0043] Figure 3 This is a schematic structural diagram of a preferred embodiment of a device for generating vehicle driving data provided by the present invention;

[0044] Figure 4 It is a structural diagram of a preferred embodiment of a terminal device provided by the present invention. DETAILED DESCRIPTION

[0045] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0046] See also Figure 1 , Figure 1 The figure is a flow chart of a preferred embodiment of a method for generating vehicle driving data provided by the present invention. The method for generating vehicle driving data comprises:

[0047] S1, obtains the original GPS data of the car traveling on each road section and the road length of each road section;

[0048] S2, dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments;

[0049] S3, performing cluster analysis on the multiple kinematic segments, classifying the multiple kinematic segments into multiple operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time based on the road speed corresponding to each operating condition type;

[0050] S4, dividing the road section into multiple road conditions according to the proportion of kinematic segments of various working condition types in the data of each road section, and calculating the displacement generated by the kinematic segments of various working condition types in the data of each road section;

[0051] S5, training a generative adversarial network constructed by inputting the road length, the road condition, and the displacement, and obtaining position sequences of various working condition types through the trained generative adversarial network;

[0052] S6, performing transition processing on the kinematic segments in the position sequence to obtain driving data of the vehicle under specific road conditions.

[0053] Specifically, this embodiment first obtains raw GPS data for a vehicle traveling on multiple different road sections, along with the road length of each section. The obtained raw GPS data is then divided into kinematic segments to produce multiple kinematic segments. Next, cluster analysis is performed on the multiple kinematic segments, classifying them into various operating condition types. The kinematic segments for each operating condition type are then fitted as a function of the corresponding road speed over time, based on the road speed corresponding to the kinematic segments for each operating condition type. The road sections are then classified into various road conditions based on the proportion of kinematic segments of each operating condition type within the data for each road section. The displacements generated by the kinematic segments for each operating condition type within the data for each road section are then calculated. A generative adversarial network constructed using the road length, road condition, and displacement as inputs is then trained to produce a trained generative adversarial network. This trained generative adversarial network can then generate a position sequence containing kinematic segments for each operating condition type. For example, assuming that constant speed driving, overtaking, avoiding, and traffic lights correspond to 1, 2, 3, and 4, respectively, the generated position sequence is 123211143. Since only the position sequence of each kinematic segment is generated and there is no connection between each kinematic segment, it is necessary to perform transition processing on the kinematic segments in the position sequence to obtain complete driving data of the car under specific road conditions.

[0054] This example trains a generative adversarial network (GAN) by selecting real-world driving data from different road sections and road length features, leveraging its adversarial training properties to achieve superior driving data generation. During the data processing phase, road condition features are extracted from each data point and combined with road length, speed, and conditions to generate vehicle driving data. This data more closely resembles real-world driving conditions, effectively improving the authenticity and accuracy of the generated data. Furthermore, the GAN can flexibly adjust the input road length, speed, and conditions to meet varying data requirements, generating driving data for vehicles under specific road conditions.

[0055] In another preferred embodiment, the division of the original GPS data into kinematic segments in S2 is specifically as follows:

[0056] All extreme points in the original GPS data are extracted, and kinematic segments are divided according to the extreme points.

[0057] Specifically, this embodiment extracts all extreme points from the acquired raw GPS data and divides the kinematic segments based on these extreme points. For example, this embodiment starts with the first extreme point at the starting point, and every three extreme points form a kinematic segment. If the velocity at the second and third extreme points is zero, a fourth extreme point is added as a kinematic segment to effectively divide the kinematic segments.

[0058] In another preferred embodiment, after dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments, the step S2 further includes:

[0059] calculating the acceleration of each kinematic segment and determining whether there is abnormal acceleration;

[0060] If abnormal acceleration exists, the kinematic segment corresponding to the abnormal acceleration is removed.

[0061] Specifically, after dividing the raw GPS data into multiple kinematic segments, this embodiment further calculates the acceleration of each kinematic segment and determines whether any abnormal acceleration exists. If any abnormal acceleration exists, the kinematic segment corresponding to the abnormal acceleration is removed to ensure data accuracy.

[0062] In another preferred embodiment, the step S3 of performing cluster analysis on the plurality of kinematic segments, classifying the plurality of kinematic segments into a plurality of operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time in combination with the road speed corresponding to each operating condition type, specifically includes:

[0063] S301, performing cluster analysis on the multiple kinematic segments using a K-means algorithm to classify the multiple kinematic segments into multiple operating condition types; wherein the operating condition types include at least uniform speed driving, overtaking, avoiding, and traffic lights;

[0064] S302 , selecting a classification center point for each operating condition type and N points closest to the classification center point, and fitting the kinematic segment of each operating condition type into a function of road speed with respect to time in combination with the road speed corresponding to each operating condition type; wherein N is a positive integer.

[0065] Specifically, this embodiment uses the K-means algorithm to perform cluster analysis on multiple kinematic segments and classify the multiple kinematic segments into multiple working condition types. For example, see Figure 2 , Figure 2This is an example diagram of the operating condition types of a kinematic segment in a method for generating vehicle driving data provided by the present invention. The operating condition types of the kinematic segment include at least constant speed driving, overtaking, avoiding, and traffic lights. Then, a classification center point for each operating condition type and the N points closest to the classification center point are selected. The kinematic segment for each operating condition type is fitted as a function of road speed with respect to time, taking into account the road speed corresponding to each operating condition type. In this embodiment, N is a positive integer, and N is preferably 10, meaning that the classification center point for each operating condition type and the 10 points closest to the classification center point are selected.

[0066] It should be noted that the algorithm used in this embodiment for cluster analysis of kinematic segments is not limited to the K-means algorithm, and other algorithms that can achieve the cluster analysis effect of this embodiment are within the scope of protection of the present invention.

[0067] This embodiment fits the kinematic segment of each operating condition type to a function of road speed with respect to time, thereby obtaining the influence of road speed on the kinematic segment. Thus, the kinematic segment corresponding to the road speed can be determined by inputting the road speed.

[0068] In another preferred embodiment, the step S4, dividing the road section into multiple road conditions according to the proportion of kinematic segments of various working condition types in each road section data, and calculating the displacement generated by the kinematic segments of various working condition types in each road section data, specifically includes:

[0069] S401, calculating the proportion of kinematic segments of various working condition types in each road section data;

[0070] S402, using a K-means algorithm to classify the road segment into multiple road conditions according to the ratio; wherein the road conditions include at least smooth, slow, congested, and severely congested;

[0071] S403: Calculate the displacements generated by kinematic segments of various working condition types in each road section data.

[0072] Specifically, this embodiment first calculates the proportion of kinematic segments of various operating conditions within each road segment's data. For example, the data for a particular road segment may contain 80% uniform speed segments, 8% overtaking segments, 10% avoidance segments, and 2% traffic light segments. Then, a K-means algorithm is used to classify the multiple road segments into various road conditions based on the proportion of kinematic segments of various operating conditions within each road segment's data. Road conditions include at least smooth, slow-moving, congested, and severely congested. Finally, since the kinematic segments corresponding to the road speed can be determined simply by inputting the road speed, the displacements generated by the kinematic segments of various operating conditions within each road segment's data can be calculated.

[0073] It should be noted that the algorithm used in this embodiment when performing cluster analysis on the proportion of kinematic segments of various working conditions in each road section data is not limited to the K-means algorithm. Other algorithms that can achieve the cluster analysis effect of this embodiment are within the scope of protection of the present invention.

[0074] In another preferred embodiment, the training method of the generative adversarial network specifically includes:

[0075] Inputting the road length, the road condition, and the displacement into a generator in a constructed generative adversarial network, the generator generating an initial position sequence of various working condition types;

[0076] Inputting the initial position sequence into a discriminator in the generative adversarial network, and allowing the discriminator to determine whether the initial position sequence conforms to actual driving conditions;

[0077] The generator and the discriminator are iteratively updated according to the judgment result until the discriminator determines that the initial position sequence generated by the generator conforms to the actual driving situation, thereby obtaining a trained generative adversarial network.

[0078] Specifically, Generative Adversarial Networks (GAN) mainly consists of two parts, namely the generator and the discriminator. When using a generative adversarial network to generate data, the generator is mainly used to learn real data so that the data it generates is more real in order to deceive the discriminator. The discriminator needs to distinguish the true from the false of the received data. Throughout the process, the generator strives to make the generated data more real, while the discriminator strives to identify the true from the false. This process is equivalent to a two-person game. As time goes by, the generator and the discriminator are constantly competing, and eventually the two networks reach a dynamic equilibrium: the data generated by the generator is close to the real data distribution, while the discriminator cannot identify the true from the false. In this embodiment, when training the generative adversarial network, the road length, road condition and displacement are input into the generator in the generative adversarial network constructed, and the generator generates an initial position sequence of various working conditions, and the initial position sequence is input into the discriminator in the generative adversarial network. Based on the features input from the generator, the discriminator determines whether the generated trajectory features match the features input to the generator (including road length, road speed, and road conditions), thereby determining whether the position sequence generated by the generator is consistent with actual driving conditions. The generator and discriminator are then iteratively updated based on the judgment results until the discriminator determines that the initial position sequence generated by the generator is consistent with actual driving conditions, resulting in a trained generative adversarial network.

[0079] This example trains a generative adversarial network using real-world driving data from different road sections and road length characteristics, leveraging its adversarial training properties to achieve superior driving data generation. During the data processing phase, road condition characteristics are extracted from each data point and combined with road length, road speed, and road conditions to generate vehicle driving data. This data more closely resembles real-world driving conditions, effectively improving the authenticity and accuracy of the generated data. Furthermore, the generative adversarial network can flexibly adjust the input road length, road speed, and road conditions to meet varying data requirements, generating driving data for vehicles under specific road conditions.

[0080] In another preferred embodiment, the step S6 of performing transition processing on the kinematic segments in the position sequence to obtain driving data of the vehicle under specific road conditions specifically includes:

[0081] S601, extracting two node data before and after the connection point of any two adjacent kinematic segments in the position sequence;

[0082] S602: Newton interpolation method is used to supplement the intermediate data between the two node data, so as to achieve a smooth transition between any two adjacent kinematic segments, thereby obtaining driving data of the vehicle under specific road conditions.

[0083] Specifically, since the trained generative adversarial network only generates position sequences of individual kinematic segments, there is no connection between each kinematic segment. For example, assuming that uniform speed driving, overtaking, avoiding, and traffic lights correspond to 1, 2, 3, and 4 respectively, the generated position sequence is 123211143. Therefore, it is necessary to perform transition processing on the kinematic segments in the position sequence. First, the two node data before and after the connection point of any two adjacent kinematic segments in the position sequence are extracted. Then, the intermediate data between the two node data is supplemented using the Newton interpolation method to ensure a smooth transition between any two adjacent kinematic segments, thereby obtaining complete driving data of the car under specific road conditions.

[0084] Correspondingly, the present invention also provides a device for generating automobile driving data, which can implement all the processes of the method for generating automobile driving data in the above embodiment.

[0085] See also Figure 3 , Figure 3 The figure is a schematic structural diagram of a preferred embodiment of a vehicle driving data generating device provided by the present invention. The vehicle driving data generating device comprises:

[0086] An acquisition module 301 is used to acquire original GPS data of the vehicle traveling on each road section and the road length of each road section;

[0087] A kinematic segment division module 302 is configured to divide the original GPS data into kinematic segments to obtain a plurality of kinematic segments;

[0088] a first clustering module 303 for performing cluster analysis on the plurality of kinematic segments, classifying the plurality of kinematic segments into a plurality of operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time based on a road speed corresponding to each operating condition type;

[0089] The second clustering module 304 is configured to classify the road section into multiple road conditions according to the proportion of kinematic segments of each operating condition type in each road section data, and calculate the displacement generated by the kinematic segments of each operating condition type in each road section data;

[0090] A training module 305 is configured to train a generative adversarial network constructed by inputting the road length, the road condition, and the displacement, and obtain position sequences of various working condition types through the trained generative adversarial network;

[0091] The transition processing module 306 is used to perform transition processing on the kinematic segments in the position sequence to obtain driving data of the vehicle under specific road conditions.

[0092] Preferably, the kinematic segment division module 302 is specifically configured to:

[0093] All extreme points in the original GPS data are extracted, and kinematic segments are divided according to the extreme points.

[0094] Preferably, the kinematic segment division module 302 is further configured to:

[0095] calculating the acceleration of each kinematic segment and determining whether there is abnormal acceleration;

[0096] If abnormal acceleration exists, the kinematic segment corresponding to the abnormal acceleration is removed.

[0097] Preferably, the first clustering module 303 specifically includes:

[0098] A first clustering unit 313 is configured to perform cluster analysis on the plurality of kinematic segments using a K-means algorithm to classify the plurality of kinematic segments into a plurality of operating condition types; wherein the operating condition types include at least uniform speed driving, overtaking, avoiding, and traffic lights;

[0099] The fitting unit 323 is configured to select a classification center point for each operating condition type and the N points closest to the classification center point, and fit the kinematic segment of each operating condition type into a function of the road speed with respect to time in combination with the road speed corresponding to each operating condition type; wherein N is a positive integer.

[0100] Preferably, the second clustering module 304 specifically includes:

[0101] A ratio calculation unit 314 is used to calculate the ratio of kinematic segments of various working condition types in each road segment data;

[0102] A second clustering unit 324 is configured to use a K-means algorithm to classify the road segments into multiple road conditions according to the ratios; wherein the road conditions include at least smooth, slow, congested, and severely congested;

[0103] The displacement calculation unit 334 is used to calculate the displacement generated by the kinematic segments of various working condition types in each road segment data.

[0104] Preferably, the training method of the generative adversarial network specifically includes:

[0105] Inputting the road length, the road condition, and the displacement into a generator in a constructed generative adversarial network, the generator generating an initial position sequence of various working condition types;

[0106] Inputting the initial position sequence into a discriminator in the generative adversarial network, and allowing the discriminator to determine whether the initial position sequence conforms to actual driving conditions;

[0107] The generator and the discriminator are iteratively updated according to the judgment result until the discriminator determines that the initial position sequence generated by the generator conforms to the actual driving situation, thereby obtaining a trained generative adversarial network.

[0108] Preferably, the transition processing module 306 specifically includes:

[0109] An extraction unit 316 is used to extract two node data before and after the connection point of any two adjacent kinematic segments in the position sequence;

[0110] The transition unit 326 is used to use Newton interpolation to supplement the intermediate data between the two node data to achieve a smooth transition between any two adjacent kinematic segments, thereby obtaining the driving data of the vehicle under specific road conditions.

[0111] In specific implementation, the working principle, control process and technical effects achieved by the vehicle driving data generation device provided in the embodiment of the present invention are the same as those of the vehicle driving data generation method in the above embodiment, and will not be repeated here.

[0112] See also Figure 4 , Figure 44 is a schematic diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 401, a memory 402, and a computer program stored in the memory 402 and configured to be executed by the processor 401. When the processor 401 executes the computer program, it implements the method for generating vehicle driving data described in any of the above embodiments.

[0113] Preferably, the computer program can be divided into one or more modules / units (e.g., computer program 1, computer program 2, ...), which are stored in the memory 402 and executed by the processor 401 to implement the present invention. The one or more modules / units can be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0114] The processor 401 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor, or the processor 401 can be any conventional processor. The processor 401 is the control center of the terminal device, and uses various interfaces and lines to connect the various parts of the terminal device.

[0115] The memory 402 mainly includes a program storage area and a data storage area. The program storage area can store an operating system, at least one application required for a function, and the data storage area can store related data. In addition, the memory 402 can be a high-speed random access memory or a non-volatile memory, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, and a Flash Card. Alternatively, the memory 402 can be other volatile solid-state memory devices.

[0116] It should be noted that the above terminal device may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that Figure 4The structural diagram is only an example of the above-mentioned terminal device and does not constitute a limitation on the above-mentioned terminal device. It may include more or fewer components than shown in the figure, or combine certain components, or different components.

[0117] An embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for generating automobile driving data described in any of the above embodiments.

[0118] Embodiments of the present invention provide a method, apparatus, device, and storage medium for generating vehicle driving data. By selecting real driving data of vehicles on different road sections and road length characteristics to train a generative adversarial network, the adversarial training characteristics are fully utilized to achieve better driving data generation results. During the data processing stage, road condition characteristics are extracted from each piece of data, and the vehicle driving data is generated by combining road length, road speed, and road conditions. This data is closer to actual driving conditions and effectively improves the authenticity and accuracy of the generated data. Furthermore, the generative adversarial network can flexibly adjust the input road length, road speed, and road conditions according to different data requirements to generate vehicle driving data under specific road conditions.

[0119] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0120] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for generating automobile driving data, characterized in that: include: Obtain the original GPS data of the car traveling on each road segment and the road length of each road segment; Dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments; performing a cluster analysis on the plurality of kinematic segments, classifying the plurality of kinematic segments into a plurality of operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time in combination with a road speed corresponding to each operating condition type; According to the proportion of kinematic segments of various working conditions in each road segment data, the road segment is divided into multiple road conditions, and the displacement generated by the kinematic segments of various working conditions in each road segment data is calculated; Training a generative adversarial network constructed using the road length, the road condition, and the displacement inputs, and obtaining position sequences of various working condition types through the trained generative adversarial network; The kinematic segments in the position sequence are subjected to transition processing to obtain driving data of the vehicle under specific road conditions.

2. The method for generating automobile driving data according to claim 1, wherein: The kinematic segment division of the original GPS data is specifically as follows: All extreme points in the original GPS data are extracted, and kinematic segments are divided according to the extreme points.

3. The method for generating automobile driving data according to claim 2, wherein: After dividing the original GPS data into kinematic segments to obtain a plurality of kinematic segments, the method further includes: calculating the acceleration of each kinematic segment and determining whether there is abnormal acceleration; If abnormal acceleration exists, the kinematic segment corresponding to the abnormal acceleration is removed.

4. The method for generating automobile driving data according to claim 1, wherein: The performing cluster analysis on the multiple kinematic segments, classifying the multiple kinematic segments into multiple operating condition types, and fitting the kinematic segments of each operating condition type into a function of road speed with respect to time based on the road speed corresponding to each operating condition type, specifically includes: Performing cluster analysis on the multiple kinematic segments using a K-means algorithm to classify the multiple kinematic segments into multiple operating condition types; wherein the operating condition types include at least constant speed driving, overtaking, avoiding, and traffic lights; The classification center point of each operating condition type and the N points closest to the classification center point are selected, and the kinematic segment of each operating condition type is fitted into a function of road speed with respect to time in combination with the road speed corresponding to each operating condition type; where N is a positive integer.

5. The method for generating automobile driving data according to claim 1, wherein: The method of dividing a road section into multiple road conditions according to the proportion of kinematic segments of various working condition types in each road section data and calculating the displacement generated by the kinematic segments of various working condition types in each road section data specifically includes: Calculate the proportion of kinematic segments of various working condition types in each road segment data; Using a K-means algorithm to classify the road section into multiple road conditions according to the ratio; wherein the road conditions include at least smooth, slow, congested, and severely congested; Calculate the displacements generated by the kinematic segments of various operating conditions in each road segment data.

6. The method for generating automobile driving data according to any one of claims 1 to 5, characterized in that: The training method of the generative adversarial network specifically includes: Inputting the road length, the road condition, and the displacement into a generator in a constructed generative adversarial network, the generator generating an initial position sequence of various working condition types; Inputting the initial position sequence into a discriminator in the generative adversarial network, and allowing the discriminator to determine whether the initial position sequence conforms to actual driving conditions; The generator and the discriminator are iteratively updated according to the judgment result until the discriminator determines that the initial position sequence generated by the generator conforms to the actual driving situation, thereby obtaining a trained generative adversarial network.

7. The method for generating automobile driving data according to claim 6, wherein: The kinematic segments in the position sequence are subjected to transition processing to obtain driving data of the vehicle under specific road conditions, specifically including: Extracting two node data before and after the connection point of any two adjacent kinematic segments in the position sequence; Newton interpolation is used to supplement the intermediate data between the two node data to achieve a smooth transition between any two adjacent kinematic segments, thereby obtaining the driving data of the car under specific road conditions.

8. A device for generating automobile driving data, characterized in that: include: An acquisition module is used to obtain the original GPS data of the car traveling on each road section and the road length of each road section; A kinematic segment division module is used to divide the original GPS data into kinematic segments to obtain a plurality of kinematic segments; a first clustering module, configured to perform cluster analysis on the plurality of kinematic segments, classify the plurality of kinematic segments into a plurality of operating condition types, and fit the kinematic segments of each operating condition type into a function of road speed with respect to time based on a road speed corresponding to each operating condition type; The second clustering module is used to classify the road section into multiple road conditions according to the proportion of kinematic segments of various working condition types in the data of each road section, and calculate the displacement generated by the kinematic segments of various working condition types in the data of each road section; a training module, configured to train a generative adversarial network constructed by inputting the road length, the road condition, and the displacement, and obtain position sequences of various working condition types through the trained generative adversarial network; The transition processing module is used to perform transition processing on the kinematic segments in the position sequence to obtain driving data of the vehicle under specific road conditions.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for generating automobile driving data according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for generating automobile driving data according to any one of claims 1 to 7.

Images