Driving condition construction method and device

By obtaining and clustering vehicle driving information, building a driving condition curve that is suitable for specific scenarios, solving the problem of deviation in test results of existing standards in specific scenarios, and achieving more accurate energy consumption and performance analysis.

CN118918555APending Publication Date: 2024-11-08JINGWEI HIRAIN (TIANJIN) RES&DEV CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411055698.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

When the existing vehicle driving conditions standards are used in specific driving scenarios, the results of energy consumption testing and performance calibration have a large deviation from the actual situation, and cannot accurately reflect the vehicle's movement characteristics in different scenarios.

Method used

By obtaining the driving information of the vehicle in the driving scenario to be tested, dividing and clustering itinerary segments, building a motion curve, and using the recurrent neural network model to generate a driving condition curve that is adapted to the scene, including the relationship between speed and time and vehicle steering The relationship between angles forms the driving conditions that match the scene.

Benefits of technology

It has achieved the construction of accurate driving conditions standards in specific driving scenarios, improved the accuracy of vehicle energy consumption testing and performance analysis, and ensured that the test results are in line with the actual situation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918555A_ABST
    Figure CN118918555A_ABST
Patent Text Reader

Abstract

The invention provides a driving condition construction method and device, and relates to the technical field of automobiles, and the method comprises the steps: obtaining the driving information of a vehicle in the operation process in a to-be-detected driving scene; dividing the running process into a plurality of first travel segments according to the driving information; clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition; for any first driving working condition in the multiple driving working conditions, determining a motion curve corresponding to the first driving working condition according to a preset vehicle speed interval and vehicle speeds in multiple second travel segments in the first driving working condition; the motion curves corresponding to all the first driving working conditions are spliced, and a first driving working condition curve is obtained; and inputting the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve. By adopting the steps, the accuracy of the results of energy consumption test and performance calibration under the vehicle driving condition can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of automobile technology, and in particular to a method and device for constructing a driving condition. Background Art

[0002] The vehicle driving condition is also called the vehicle operating cycle, which is a speed-time curve generally described as the driving characteristics of a certain type of vehicle (such as passenger cars, buses, heavy vehicles, etc.) in a specific traffic environment (such as highways, urban roads, ports, etc.). It can characterize the kinematic characteristics of a car in a specific scenario, and is the basis for energy consumption testing and performance analysis of the vehicle, and is also the main benchmark for calibration and optimization of various performance indicators of the vehicle.

[0003] At present, all countries with developed automobile industries have formulated their own vehicle driving condition standards based on their own automobile driving condition environments, such as the FTP75 standard in the United States, the NEDC and WLTC standards in Europe, the JAPAN10-15 in Japan, and the CLTC standard in China.

[0004] The above vehicle driving condition standards are all formulated based on driving data from urban roads and highways in various countries, and they mainly reflect the driving conditions and driving habits of passenger cars and commercial vehicles on public roads in the region. For specific driving scenarios, such as industrial parks, port terminals, and closed-loop passenger routes, the driving characteristics of vehicles (such as average speed, maximum speed, and acceleration) are significantly different from those of public road vehicles. Therefore, when using vehicle driving condition standards based on public road data to conduct energy consumption tests and performance calibrations in specific driving scenarios, the development and test results will deviate significantly from the actual situation. Summary of the invention

[0005] A driving condition construction method and device provided in the present application can improve the accuracy of the results of energy consumption testing and performance calibration under vehicle driving conditions.

[0006] In a first aspect, an embodiment of the present application provides a driving condition construction method, the method comprising:

[0007] In the driving scenario to be tested, obtain the driving information of the vehicle during operation;

[0008] Dividing the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a driving process of the vehicle between two adjacent parking states;

[0009] Clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition;

[0010] For any first driving condition among the plurality of driving conditions, determining a motion curve corresponding to the first driving condition according to a preset vehicle speed interval and vehicle speeds in a plurality of second travel segments in the first driving condition;

[0011] The motion curves corresponding to each first driving condition are spliced ​​to obtain a first driving condition curve, where the first driving condition curve is used to represent the relationship between the speed and time of the vehicle in the driving scenario to be tested.

[0012] In a second aspect, the present application provides a driving condition construction device, the device comprising:

[0013] An acquisition module, used to acquire driving information of the vehicle during operation in the driving scenario to be tested;

[0014] a dividing module, configured to divide the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a running process of the vehicle between two adjacent parking states;

[0015] A clustering module, used for clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition;

[0016] An establishing module is used to determine, for any first driving condition among the plurality of driving conditions, a motion curve corresponding to the first driving condition according to a preset vehicle speed interval and vehicle speeds in a plurality of second travel segments in the first driving condition;

[0017] The splicing module is used to splice the motion curves corresponding to each first driving condition to obtain a first driving condition curve, wherein the first driving condition curve is used to represent the relationship between the speed and time of the vehicle during operation.

[0018] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0019] When the processor executes the computer program instructions, it implements the driving condition construction method as described in any one of the embodiments of the first aspect.

[0020] In a fourth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, a driving condition construction method as in any one of the embodiments in the first aspect is implemented.

[0021] In a fifth aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes a driving condition construction method as in any one of the embodiments in the first aspect above.

[0022] In the driving condition construction method and device of the embodiment of the present application, the method includes: in the driving scene to be tested, obtaining driving information of the vehicle during operation; dividing the operation process into multiple first travel segments according to the driving information, the first travel segment is the driving process of the vehicle between two adjacent parking states; clustering the multiple first travel segments to obtain multiple driving conditions, and multiple second travel segments included in each driving condition; for any first driving condition in the multiple driving conditions, determining the motion curve corresponding to the first driving condition according to the preset vehicle speed range and the vehicle speed in the multiple second travel segments in the first driving condition; splicing the motion curves corresponding to each first driving condition to obtain a first driving condition curve, the first driving condition curve is used to represent the relationship between the speed and time of the vehicle in the driving scene to be tested; inputting the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve, the second driving condition curve is used to represent the vehicle steering angle corresponding to the vehicle at different times and speeds, and the recurrent neural network model is trained using the driving information. In this embodiment, through the above steps, the driving information of the vehicle in the driving scene to be tested can be obtained, the motion curve of the vehicle under each driving condition can be analyzed, and finally the motion curve under each driving condition can be spliced ​​into a first driving condition curve representing the relationship between the vehicle and the speed, and based on the first driving condition curve, a second driving condition curve for representing the wheel steering angle at different times and vehicle speeds can be determined, so that the driving condition of the vehicle in the driving scene can be constructed, and then the driving condition matching the driving scene can be constructed in a targeted manner for the driving scene that does not match the current general driving condition standard. This allows the input of driving condition standards that meet the actual situation when performing vehicle energy consumption testing and performance analysis for the driving scene. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] Figure 1 It is a flowchart of a driving condition construction method provided by an embodiment of the present application;

[0025] Figure 2It is a schematic diagram of a flow chart of determining a motion curve in a driving condition construction method provided in an embodiment of the present application;

[0026] Figure 3 This is a structural diagram of an LSTM model in a driving condition construction method provided in an embodiment of the present application;

[0027] Figure 4 This is a schematic diagram of LSTM model neurons in a driving condition construction method provided in one embodiment of the present application;

[0028] Figure 5 A schematic diagram of the overall process of constructing a second driving condition curve in a driving condition construction method provided in one embodiment of the present application is shown;

[0029] Figure 6 A schematic diagram of the overall process of constructing a first driving condition curve in a driving condition construction method provided in one embodiment of the present application is shown;

[0030] Figure 7 It is a structural schematic diagram of a driving condition construction device provided in an embodiment of the present application;

[0031] Figure 8 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0033] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0034] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0035] In order to solve the problems of the prior art, the embodiment of the present application provides a driving condition construction method and device. The driving condition construction method provided by the embodiment of the present application is first introduced below.

[0036] Figure 1 FIG. 1 is a flow chart of a driving condition construction method provided by an embodiment of the present application. Figure 1 As shown, the method may specifically include the following steps:

[0037] Step 101, in a driving scenario to be tested, obtaining driving information of a vehicle during operation;

[0038] When the vehicle is running in the driving scenario to be tested, the collector can be used to collect the driving information required for the vehicle to run through the vehicle CAN bus, and the driving information can be sent to the big data cloud platform through the vehicle T-Box. The big data cloud platform stores the driving information in the relevant server, and data analysts obtain the driving information through the big data platform for subsequent processing.

[0039] The above-mentioned driving scenarios to be tested can be industrial park scenarios, port terminal scenarios, closed-loop passenger route scenarios, etc., and those skilled in the art can select scenarios for energy consumption testing and performance calibration as needed. The above-mentioned driving information may include a vehicle speed-time data set and a wheel turning angle-vehicle speed data set. The driving information can be obtained by obtaining the vehicle speed and vehicle steering wheel angle data during operation. The above data are arranged in time series, and the data update frequency is set to 1Hz through the resampling method, that is, the data is updated once per second. Then the collected data is analyzed, and the abnormal data generated during the equipment collection and data transmission process is filtered and eliminated to obtain the above-mentioned vehicle speed-time data set and wheel turning angle-vehicle speed data set.

[0040] Step 102, dividing the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a running process of the vehicle between two adjacent parking states;

[0041] The actual driving of vehicles is very random, and it is difficult to directly perform regular statistics and feature calculations on them. Therefore, it is necessary to process the original data and divide it into periodic driving segments, that is, to divide the long-term driving data into multiple short-trip segments, namely the first trip segments. The first trip segment refers to the driving process between two adjacent parking states, that is, a single motion cycle of the vehicle from static to moving and then to static. The vehicle contains a large number of deceleration, acceleration, and uniform speed motion states in this motion cycle.

[0042] In the actual segmentation process, considering that there are a lot of long-term parking situations during vehicle operation, in order to avoid long-term parking of vehicles resulting in too high a proportion of idle time in the first trip segment, which leads to distortion of slice features and affects subsequent calculations, the following segmentation criteria for the first trip segment can be determined:

[0043] 1. When the idling time before a certain first travel segment enters the motion state (speed greater than 0) exceeds 30 seconds, the idling time is cut into 30-second cycles and is not included in the first travel segment.

[0044] 2. When the idling time before a first trip segment enters the motion state accounts for more than 50% of the entire first trip segment slice, 30% of the idling state time of the entire segment needs to be cut.

[0045] 3. When a first travel segment enters the moving state and then enters the idle state again, if it remains in the idle state for more than 10 seconds, it is considered that a short travel cycle is completed, and the subsequent driving state is divided into the next first travel segment; if the idle state is maintained for no more than 10 seconds and the vehicle enters the moving state again, it is determined that the vehicle is still in the same first travel segment.

[0046] Step 103, clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition;

[0047] In this embodiment, after completing the short-trip slicing of the running process, it is necessary to calculate the characteristic parameters of each first trip segment. The characteristic parameters need to be able to reflect the overall distribution of the vehicle's motion characteristics and motion state in the segment. The 11 kinematic characteristic parameters that need to be calculated are listed in Table 1.

[0048] Serial number Characteristic parameters symbol 1 <![CDATA[Average speed (km·h -1 )]]> <![CDATA[V m ]]> 2 <![CDATA[Average driving speed (km·h -1 )]]> <![CDATA[V dm ]]> 3 <![CDATA[Maximum speed (km·h -1 )]]> <![CDATA[V max ]]> 4 <![CDATA[Standard deviation of speed (km·h -1 )]]> <![CDATA[V std ]]> 5 <![CDATA[Average acceleration in the acceleration phase (m·s -2 )]]> <![CDATA[a ac ]]> 6 <![CDATA[Average deceleration in the deceleration phase (m·s -2 )]]> <![CDATA[a de ]]> 7 <![CDATA[Standard deviation of acceleration (m·s -2 )]]> <![CDATA[a std ]]> 8 Idle time ratio (%) <![CDATA[p i ]]> 9 Acceleration state time ratio (%) <![CDATA[p a ]]> 10 Deceleration time ratio (%) <![CDATA[p d ]]> 11 Time ratio of uniform speed state (%) <![CDATA[p c ]]>

[0049] Table 1, kinematic characteristic parameter table

[0050] After calculating the 11 characteristic parameters of each first trip segment, the first trip segment characteristic database can be obtained. Then, the key characteristic parameters are screened from the first trip segment characteristic database. Finally, according to the key characteristic parameters, various operating conditions and the motion curve corresponding to each driving condition are obtained.

[0051] Specifically, in one embodiment, step 103 further includes:

[0052] Analyze the multiple first characteristic parameter values ​​included in each first trip segment to obtain key characteristic parameters;

[0053] Constructing a score matrix corresponding to each of the first trip segments according to the key feature parameters;

[0054] Clustering is performed based on the score matrix corresponding to each of the first travel segments to obtain a plurality of second travel segments under a variety of driving conditions.

[0055] In this embodiment, after obtaining the first trip segment feature database, that is, multiple first feature parameter values, principal component analysis (PCA) is required to perform dimensionality reduction processing to obtain each principal component and its corresponding contribution rate, so as to determine the principal component with the highest contribution rate for the next step of operating condition clustering.

[0056] It should be noted that before the principal component analysis, because the dimensions of each characteristic parameter are different and the numerical differences are large, it is necessary to perform standardization to eliminate the differences in the dimensions of each characteristic parameter and improve the accuracy of the PCA dimension reduction results. The Z-score standardization method is mainly used. The data after standardization is subjected to PCA analysis, and the number of principal components generated can be set to 7 to 8 based on actual experience. The reason is that generally speaking, the contribution rate of the principal components ranked after 7 to the whole is almost negligible.

[0057] After PCA analysis, the contribution rate of each principal component is sorted from high to low, and the number of principal components required to be input in the next step is determined according to the following criteria: when the cumulative contribution rate of the current N principal components is greater than 85%, it indicates that these N principal components can characterize the characteristic parameters of the first trip segment feature database in most cases. Therefore, the first N principal components are selected to construct the principal component score matrix for the next step of cluster analysis.

[0058] The K-means clustering algorithm is used to analyze the principal component score matrix of each short trip segment obtained in the previous stage. When setting the number of clusters of the K-means algorithm, the discrimination degree of the working conditions after clustering with different numbers is compared, and it is determined that when the number of clusters is 3, the discrimination effect of the working conditions is optimal. Therefore, the K-means clustering algorithm is used to cluster the principal component score matrix of each first trip segment, and all the first trip segments are clustered into three clusters. Each cluster represents a type of driving condition. The first trip segment assigned to a certain condition indicates that the segment belongs to the driving condition, and all the first trip segments assigned to the driving condition are second trip segments.

[0059] Step 104, for any first driving condition among the plurality of driving conditions, determining a motion curve corresponding to the first driving condition according to a preset vehicle speed range and vehicle speeds in a plurality of second travel segments in the first driving condition;

[0060] The above-mentioned vehicle speed interval can be divided into an interval every 1km / h, and of course, it can also be divided into an interval every 2km / h or 3km / h, and this embodiment does not limit this. The vehicle speed in the second travel segment can be the maximum vehicle speed in the second travel segment. The above-mentioned motion curve is a time-speed curve of the vehicle corresponding to each first driving condition. The curve can be further obtained by establishing a Markov state transition probability matrix. For details, please refer to the following embodiment, which will not be described in detail in this embodiment.

[0061] Step 105 , splicing the motion curves corresponding to each first driving condition to obtain a first driving condition curve, where the first driving condition curve is used to represent the relationship between the speed and time of the vehicle in the driving scenario to be tested.

[0062] In another embodiment, the step 105 further includes:

[0063] Obtaining a duration ratio of the second travel segment in each first driving condition;

[0064] According to the duration ratio and the preset duration corresponding to the first driving condition curve, each of the motion curves is spliced ​​to obtain the first driving condition curve.

[0065] In this embodiment, the duration ratio of the synthetic motion curve of each working condition is determined according to the proportion of the second travel segment among the three working conditions to the total duration. For example, the duration ratios of the slices of working conditions one to three are 43.75%, 20.05%, and 36.20%, respectively. When the duration of the first driving condition curve to be generated is set to 1800s, the duration of the speed curve sampled and synthesized by each of the three working conditions is set according to the above ratio. After splicing the synthetic motion curves of the three working conditions, the overall typical working condition vehicle motion curve, that is, the first driving condition curve, can be obtained.

[0066] Step 106: Input the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve, wherein the second driving condition curve is used to describe the vehicle steering angle corresponding to the vehicle at different times and speeds, and the recurrent neural network model is trained using the driving information.

[0067] In this embodiment, after obtaining a first driving condition curve under a typical driving condition of the vehicle, a second driving condition curve under a corresponding typical driving condition of the vehicle, i.e., a speed-steering angle curve, is constructed based on the first driving condition curve as an input, and the construction method includes:

[0068] Establish a long short-term memory neural network model; use a wheel angle-time data set including vehicle speed and steering wheel angle to train the long short-term memory neural network model; the model prediction error is reduced to meet the threshold requirement through training, and a long short-term memory neural network model for predicting the vehicle angle signal corresponding to the vehicle speed at a specific time is obtained; input the first driving condition speed curve, and predict the corresponding vehicle characteristic driving condition wheel angle curve. For the specific training process, please refer to the following embodiment, which will not be described in detail in this embodiment.

[0069] In this embodiment, through the above steps, the driving information of the vehicle in the driving scene to be tested can be obtained, the motion curve of the vehicle under each driving condition can be analyzed, and finally the motion curve under each driving condition can be spliced ​​into a first driving condition curve representing the relationship between the vehicle and the speed, so that the driving condition of the vehicle in the driving scene can be constructed, and then the driving condition matching the driving scene can be constructed in a targeted manner for the driving scene that does not match the current general driving condition standard. This allows the input of driving condition standards that meet the actual situation when performing vehicle energy consumption testing and performance analysis for the driving scene.

[0070] In an embodiment of the present application, for any first driving condition among the multiple driving conditions, determining a motion curve corresponding to the first driving condition according to a preset vehicle speed range and vehicle speeds in a plurality of second travel segments in the first driving condition includes:

[0071] For any first driving condition among the plurality of driving conditions, establishing a state transition probability matrix corresponding to the first driving condition, wherein the state transition probability matrix is ​​used to represent the probability of the vehicle changing between different driving states under the first driving condition;

[0072] The state transition probability matrix corresponding to the first driving condition is sampled to determine a motion curve corresponding to the first driving condition.

[0073] In this embodiment, after K-means clustering, a corresponding probability transfer matrix is ​​established for each operating condition according to the classified operating conditions. This part is based on the basic principle of Markov chain, which regards the vehicle speed as a random process, divides the vehicle speed according to the interval and regards it as different driving states, and there is a causal relationship between the two adjacent driving states in time, that is, the current driving state will affect the next driving state, and the Markov chain has no memory, that is, the state at the next moment is only related to the state at the current moment, and has no correlation with the previous moments. Therefore, after converting the vehicle speed into different driving states, the driving frequency between adjacent states is counted, and the frequency is converted into the corresponding probability, and the state transfer probability matrix P can be established.

[0074] The state transition probability matrix is ​​defined as follows: Let the current state be i, the next state be j, P ij represents the probability of transitioning from the current state i to the next state j, ij is a binary state of the combination of two states. If there are K states in the state space, there are K×K transition probabilities, thus forming a probability transfer matrix P ij (i, j = 1, 2, ..., K), the matrix can be expressed as formula (1), the row coordinate of each element of the matrix represents the state at the current moment, and the column coordinate represents the state at the next moment. Therefore, the probability of falling into a certain state interval at the next moment can be known from the transition probability matrix, so as to select the state of the sampling point at the next moment.

[0075]

[0076] In this embodiment, by establishing a state transition probability matrix for different driving conditions, the transition rules of the vehicle driving state under different conditions can be reflected. This makes the determination of the motion curve closer to the actual driving situation and can more accurately reflect the motion characteristics of the vehicle under different driving conditions.

[0077] In one embodiment of the present application, a state transition probability matrix corresponding to the first driving condition is established according to a preset vehicle speed range and the vehicle speeds in a plurality of second travel segments in the first driving condition, including:

[0078] In the vehicle speed interval, determining a target interval corresponding to the vehicle speed in each of a plurality of second travel segments in the first driving condition;

[0079] According to a preset correspondence between the target interval and the driving state, obtaining the driving state of the vehicle in the plurality of second travel segments;

[0080] Obtain the driving frequency between driving states at adjacent times;

[0081] Constructing a state transition frequency matrix corresponding to each driving condition according to the driving frequency;

[0082] The state transition probability matrix corresponding to each driving condition is established according to the state transition frequency matrix.

[0083] In this embodiment, the vehicle speed in the second travel segment is the highest vehicle speed in the second travel segment. Each vehicle speed is included in the vehicle speed interval, and the vehicle speed interval can be divided into an interval of 1 km / h. Assuming that the vehicle speed in the second travel segment is 3 km / h, the second travel segment belongs to the 3-4 km / h interval, and the 3-4 km / h interval is the target interval.

[0084] Each vehicle speed interval corresponds to a driving state. After the target interval is determined, the driving state of the target interval, that is, the driving state of the second travel segment, can be determined directly according to the corresponding relationship.

[0085] For each working condition, all driving states are traversed, and the transition frequency relationship between different motion states of all second trip segments under the working condition is counted. For example, in a certain working condition, the vehicle speed changes from 5 km / h at the previous moment to 6 km / h at the current moment, then the element P corresponding to the state space matrix 56 The corresponding value is added by 1. The driving states of all the second travel segments under the working condition are traversed, and the state transitions at each moment are counted according to the corresponding speed-motion state relationship, and the frequencies of the corresponding elements are accumulated in the state space, and finally the state transition frequency matrix of the working condition is obtained.

[0086] After obtaining the state transition frequency matrix, the corresponding state transition probability matrix is ​​calculated by formula (2), where N ij is the travel frequency from state i to state j, P ij represents the probability of state i transitioning to state j, and D represents the total number of motion states.

[0087]

[0088] After calculation by formula (2), the state transition probability matrix of each working condition can be obtained.

[0089] In this embodiment, by establishing a state transition probability matrix corresponding to the driving condition, not only the data processing efficiency and accuracy are improved, but also it is helpful to explore the transition law of the vehicle driving state and provide support for the intelligent decision-making and control of the vehicle.

[0090] In one embodiment of the present application, Figure 2 FIG. 1 is a flow chart showing a process of determining a motion curve in a driving condition construction method provided in an embodiment of the present application. Figure 2 As shown, the sampling of the state transition probability matrix corresponding to the first driving condition to determine the motion curve corresponding to the first driving condition includes:

[0091] Step 201, searching for a target state sequence in the state transition probability matrix according to a preset first state and a preset random number;

[0092] In one embodiment, the step 201 further includes:

[0093] Searching for an element whose driving state at the current moment is the first state in the state transition probability matrix, wherein the element represents the probability of the vehicle changing between the driving state at the current moment and the driving state at the next moment;

[0094] According to a preset random number, searching among the elements for a target element whose difference with the random number is within a preset difference;

[0095] The driving state at the next moment corresponding to the target element is determined as the target state, and the target element whose difference with the random number is within a preset difference is searched in the element according to the preset random number, until the number of the target states reaches the preset number and the last target state in the target state sequence is the preset value, thereby obtaining the target state sequence.

[0096] In this embodiment, the first state can be randomly set, usually set to 1, indicating that the speed is 0. Since each element in the state transition probability matrix represents the probability of the change between the driving state at the current moment and the next moment, the element whose driving state at the current moment is 1 can be obtained.

[0097] Since there may be multiple elements whose driving status is 1 at the current moment, a random number is used to determine the target element among the multiple elements. Specifically, the random number needs to conform to the Gaussian distribution, and the number of random numbers is determined by the first driving condition curve to be generated. If the relationship between the random number x and any element among the multiple elements conforms to the following formula:

[0098]

[0099] in, and Both indicate that the state transition probability matrix includes elements of the first state.

[0100] The relationship between the random number x and any element in the multiple elements conforms to the above formula, then the corresponding P ij is the target element, and k is the driving state at the next moment, that is, the target state.

[0101] After determining the next driving state, the above process is repeated to obtain a set of driving state sequences of a preset number, and the sequence needs to satisfy that the last state returns to the initial state 1 again, that is, the speed returns to 0.

[0102] Step 202, converting the standard state sequence into an initial speed curve;

[0103] After obtaining the target state sequence, the target state sequence is converted into an initial speed curve by the following formula: the initial speed curve includes a speed value corresponding to each of the target states;

[0104] The speed value v k Calculate using the following expression:

[0105] v k =[(k-1)+x]·Δd (4)

[0106] Wherein, x is a random number, Δd is the length of the target state sequence, and k is the value corresponding to the target state.

[0107] Step 203: When the difference between the characteristic parameter error of the initial speed curve and the characteristic parameter error of the first driving condition is within a preset threshold, the initial speed curve is determined as the motion curve.

[0108] In this embodiment, if the mean absolute error value is within 5%, it is regarded as a qualified generated working condition that meets the statistical error requirement. Otherwise, return to step 201 to resample and construct candidate working conditions for calculation. The formula for calculating the mean absolute error is as follows, where mape is the absolute mean error, D cand and D real are the mth characteristic parameter values ​​of the first driving condition and the initial speed curve, respectively, and n is the number of characteristic parameters:

[0109]

[0110] In this embodiment, accurate control and optimization of the motion curve are achieved by introducing random numbers, characteristic parameter matching, and optimization adjustment methods.

[0111] In one embodiment of the present application, after the motion curves corresponding to each first driving condition are spliced ​​to obtain the first driving condition curve, the method further includes:

[0112] The wheel angle data is used to train a long short-term memory neural network model to obtain a recurrent neural network model;

[0113] The step of inputting the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve comprises:

[0114] The first driving condition curve and the preset vehicle weight are input into the recurrent neural network model to obtain the second driving condition curve.

[0115] In this embodiment, the driving information also includes wheel angle data, which is the wheel angle value of the vehicle at different vehicle speeds.

[0116] It should be noted that in the process of determining the second driving condition using the first driving condition curve and the preset vehicle weight, the vehicle weight input corresponding to each moment can be customized by the user so as to generate wheel angles under different vehicle weight conditions according to different vehicle weight test requirements, thereby obtaining the second driving condition curve.

[0117] Specifically, in one embodiment of the present application, the long short-term memory neural network model is trained using the wheel angle data to obtain the recurrent neural network model, including:

[0118] Inputting the wheel angle data into a long short-term memory neural network model, so that the long short-term memory neural network model performs forward propagation calculation based on a preset first model parameter to obtain a predicted value;

[0119] When a root mean square error between the predicted value and the preset actual value is greater than a target threshold, performing back propagation calculation on the long short-term memory neural network model using the predicted value to obtain a gradient value of the first model parameter;

[0120] Using the gradient value to update the first model parameter to obtain a second model parameter;

[0121] The second model parameter is determined as a new first model parameter, and the step of inputting the wheel angle data into the long short-term memory neural network model so that the long short-term memory neural network model performs forward propagation calculation based on the preset first model parameter to obtain a predicted value, until the root mean square error between the predicted value and the actual value is less than the target threshold, thereby obtaining the recurrent neural network model.

[0122] In this embodiment, the model structure diagram of the long short-term memory neural network model and the long short-term memory neural network neuron principle diagram are as shown in Figure 3 and Figure 4 As shown in Figure 1, the model structure consists of three parts: input layer, hidden layer and output layer. The input layer of the model in the figure sends the preprocessed input to the model for calculation. The input received by the input layer is represented by X (0) ; The number of hidden layers in the model is K (1, 2, ..., k, .., K), and the number of LSTM neurons in the kth layer is represented by X (k) , the connection weight connected to the hidden layer neuron is represented by W (k) , the bias weight is denoted as B (k-1) ; The model output dimension of the output layer is represented by X (K) .

[0123] When performing forward propagation calculation, the wheel angle data needs to be input into the input layer, and then the calculated result is sequentially passed through the hidden layer and the output layer to obtain the predicted value. When performing reverse propagation calculation, the predicted value needs to be sequentially passed from the output layer to the hidden layer, and then reversely calculated to the input layer to obtain the gradient value of the first model parameter. For the specific calculation method, please refer to S2 and S3 in the following embodiment, which will not be described in detail in this embodiment.

[0124] In one embodiment of the present application, the long short-term memory neural network model is trained using the wheel angle data to obtain the recurrent neural network model, including:

[0125] Normalizing the wheel angle data to obtain a first data set;

[0126] Converting the first data set into a second data set represented in vector form;

[0127] The second data set is used to train the long short-term memory neural network model to obtain the recurrent neural network model.

[0128] In this embodiment, the wheel angle data need to be processed first. The wheel angle data is a wheel angle-time data set. For the specific data processing process, please refer to S1 in the following embodiment.

[0129] Figure 5 A schematic diagram of the overall process of constructing a second driving condition curve in a driving condition construction method provided in one embodiment of the present application is shown;

[0130] S1. Data pre-processing

[0131] The wheel angle-time data set in the driving information obtained in the above embodiment is processed and classified. Specifically, the data set needs to be normalized first so that all parameters in the data set are normalized to the same order of magnitude. The normalized expression is shown in formula (1):

[0132]

[0133] After normalization, the data set is classified according to its purpose. The data in the first 80% time period of the data set is divided into a training set for training the model; the data in the last 20% time period is divided into a test set for testing the prediction accuracy of the model after the model training is completed.

[0134] Furthermore, the data set needs to be processed into vector form when it is sent to the model for training and testing. The input [vehicle speed (V), vehicle weight (W), turning angle (R)] is expressed as: [[V(t1), W(t1), R(t1-1)][V(t2), W(t2), R(t1)]…[V(tn), W(tn), R(tn-1)]]; the output data is regularized as: [[R(t 1) ][R(t2)]…[R(tn)]]; where the input of each time step tn is the vehicle speed V(tn), vehicle weight W(tn) and steering wheel angle R(tn-1) at the previous moment, and the output is the steering wheel angle R(tn) at the current moment.

[0135] S2. Build LSTM model

[0136] Construct a neural network model with a long short-term memory neural network (LSTM) model as the neuron. The model structure diagram and LSTM neuron principle diagram are shown in the following figure. Figure 3 and Figure 4 As shown in Figure 1, the model consists of three parts: input layer, hidden layer and output layer. The input layer of the model in the figure sends the preprocessed input to the model for calculation. The input received by the input layer is represented by X (0) ; The number of hidden layers in the model is K (1, 2, ..., k, .., K), and the number of LSTM neurons in the kth layer is represented by X (k) , the connection weight connected to the hidden layer neuron is represented by W (k) , the bias weight is denoted as B (k-1) ; The model output dimension of the output layer is represented by X (K) .

[0137] S3. Training LSTM model

[0138] In order to reduce the error between the real sample and the reconstructed sample during the training process, after the model is calculated by forward propagation, the network weight is updated by the back propagation algorithm. Specifically, the gradient descent method is used to optimize the neuron weight layer by layer, thereby reducing the error value. In the specific parameter optimization algorithm, because the Adam algorithm is more efficient and generalizable than other optimization algorithms, the present invention uses the Adam optimization algorithm to update the network parameters of the model. The Adam algorithm mainly estimates the learning rate of each parameter by calculating the distance of the gradient, and its expression is expressed as follows.

[0139] m t =β1*m t-1 +(1-β1)*g (7)

[0140] v t =β2*v t-1 +(1-β2)*g 2 (8)

[0141]

[0142] Among them, m t is the historical gradient V t-1 The first-order exponential smoothing value of is used to obtain the gradient value with momentum; v t is the historical gradient V t-1 The second-order exponential smoothing value is used to obtain the learning rate weight parameter of each weight parameter; lr t is the learning rate of the model, ε is a smoothing term, usually 1E-08; V t is the gradient variable value after update.

[0143] After each model training, the root mean square error between the predicted value and the actual value is calculated as the model prediction error standard. The error calculation formula is shown in formula (10):

[0144]

[0145] Where m is the length of the input dataset time series, y i and Represent the actual value and estimated value of the wheel angle at time step i respectively

[0146] When the model error during training drops to the preset target threshold, the model training stops and is saved.

[0147] S4.LSTM model testing

[0148] After the training is completed, the LSTM model obtained by training is used to predict the test set, and the error between the predicted output value and the actual output value is calculated using formula (9). In the embodiment, if the prediction error value of the test set of the model is less than 0.07, it means that the prediction accuracy of the model is acceptable, otherwise, steps S1 to S3 are repeated to retrain the model.

[0149] S5.LSTM model prediction

[0150] The first driving condition curve obtained is input into the LSTM model. In particular, the vehicle weight input corresponding to each moment can be customized by the user so as to generate wheel angles under different vehicle weight conditions according to different vehicle weight test requirements. At the same time, the initial input of the wheel angle is set to 0, that is, there is no steering action. After the model gives the current moment angle prediction value according to the initial input, the prediction value is used as the model input for the next moment to realize the rolling prediction of the model. After the vehicle typical driving condition speed curve is input into the LSTM model according to the above setting conditions, the vehicle typical driving condition steering angle curve of the corresponding duration predicted by the model can be obtained. The curve expresses the steering wheel steering angle of the vehicle at the corresponding moment and the corresponding speed, which can characterize the steering condition of the vehicle.

[0151] In this embodiment, by constructing a recurrent neural network model, the first driving condition curve and the preset vehicle weight are input into the recurrent neural network model, thereby obtaining a second driving condition curve representing the vehicle steering angle corresponding to different time and speed, so as to construct the driving condition of the vehicle in the driving scene, and then realize the construction of a driving condition matching the driving scene in a targeted manner for a driving scene that does not match the current general driving condition standard. This allows the input of a driving condition standard that meets the actual situation when performing vehicle energy consumption testing and performance analysis for the driving scene.

[0152] In one embodiment of the present application, Figure 6 A schematic diagram of the overall process of constructing a first driving condition curve in a driving condition construction method provided in one embodiment of the present application is shown;

[0153] S1, driving data collection and classification, which corresponds to obtaining driving information of the vehicle during operation in the driving scene to be tested in the above embodiment;

[0154] S2, kinematic short-trip segment division, that is, corresponding to the above-mentioned embodiment of dividing the running process into a plurality of first trip segments according to the driving information, wherein the first trip segment is the driving process of the vehicle between two adjacent parking states;

[0155] S3, calculate the kinematic characteristic parameters of short-stroke slicing;

[0156] S4, extracting key features by reducing the dimension through principal component analysis, that is, analyzing the multiple first feature parameter values ​​included in each first trip segment in the above embodiment to obtain key feature parameters;

[0157] S5, classifying the operating conditions according to the key feature K-means, that is, clustering the plurality of first travel segments in the above embodiment to obtain a plurality of driving operating conditions and a plurality of second travel segments included in each driving operating condition;

[0158] S6, calculating the state transition probability matrix of each operating condition, that is, corresponding to the above-mentioned embodiment, for any first driving condition among the multiple driving conditions, establishing a state transition probability matrix corresponding to the first driving condition;

[0159] S7, using the Monte Carlo method to sample and generate motion curves for each operating condition, that is, sampling the state transition probability matrix corresponding to the first driving condition in the above embodiment to determine the motion curve corresponding to the first driving condition;

[0160] S8, each operating condition curve is spliced ​​to generate a representative motion curve and a rationality verification is performed, that is, corresponding to the above embodiment, the motion curve corresponding to each first driving condition is spliced ​​to obtain the first driving condition curve.

[0161] Figure 7 A schematic structural diagram of a driving condition construction device provided by an embodiment of the present application is shown. For ease of explanation, only the parts related to the embodiment of the present application are shown.

[0162] Reference Figure 7 , the driving condition constructing device 700 may include:

[0163] The acquisition module 701 is used to acquire the driving information of the vehicle during operation in the driving scene to be tested;

[0164] A division module 702 is used to divide the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a driving process of the vehicle between two adjacent parking states;

[0165] A clustering module 703 is used to cluster the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition;

[0166] Establishing module 704, for determining, for any first driving condition among the plurality of driving conditions, a motion curve corresponding to the first driving condition according to a preset vehicle speed interval and vehicle speeds in a plurality of second travel segments in the first driving condition;

[0167] A splicing module 705, used for splicing the motion curves corresponding to each first driving condition to obtain a first driving condition curve, where the first driving condition curve is used to represent the relationship between the speed and time of the vehicle during operation;

[0168] Input model 706 is used to input the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve, wherein the second driving condition curve is used to express the vehicle steering angle corresponding to the vehicle at different times and speeds, and the recurrent neural network model is trained using the driving information.

[0169] Optionally, the clustering module 703 includes:

[0170] An analysis submodule, configured to analyze a plurality of first characteristic parameter values ​​included in each first trip segment to obtain key characteristic parameters;

[0171] A construction submodule, configured to construct a score matrix corresponding to each of the first trip segments according to the key feature parameters;

[0172] The clustering submodule is used to perform clustering based on the score matrix corresponding to each of the first travel segments to obtain a plurality of second travel segments under a variety of driving conditions.

[0173] Optionally, the establishing module 704 includes:

[0174] An establishment submodule is used to establish, for any first driving condition among the multiple driving conditions, a state transition probability matrix corresponding to the first driving condition, wherein the state transition probability matrix is ​​used to represent the probability of the vehicle changing between different driving states under the first driving condition;

[0175] The sampling submodule is used to sample the state transition probability matrix corresponding to the first driving condition to determine the motion curve corresponding to the first driving condition.

[0176] Optionally, the establishing submodule includes:

[0177] a first determining unit, configured to determine, in the vehicle speed interval, a target interval corresponding to a vehicle speed in each of a plurality of second travel segments in the first driving condition;

[0178] A first acquisition unit, configured to acquire the driving state of the vehicle in the plurality of second travel segments according to a preset correspondence relationship between a target interval and a driving state;

[0179] A second acquisition unit, used to acquire the driving frequency between driving states at adjacent times;

[0180] A construction unit, configured to construct a state transition frequency matrix corresponding to each driving condition according to the driving frequency;

[0181] An establishing unit is used to establish the state transition probability matrix corresponding to each driving condition according to the state transition frequency matrix.

[0182] Optionally, the sampling submodule includes:

[0183] A searching unit, configured to search for a target state sequence in the state transition probability matrix according to a preset first state and a preset random number;

[0184] A conversion unit, used for converting the standard state sequence into an initial speed curve;

[0185] The second determining unit is configured to determine the initial speed curve as the motion curve when a difference between a characteristic parameter error of the initial speed curve and a characteristic parameter error of the first driving condition is within a preset threshold.

[0186] Optionally, the searching unit includes:

[0187] A first search subunit, used for searching the state transition probability matrix for an element whose driving state at a current moment is the first state, wherein the element represents a probability of a vehicle changing between the driving state at a current moment and the driving state at a next moment;

[0188] A second search subunit is used to search, according to a preset random number, for a target element whose difference with the random number is within a preset difference value among the elements;

[0189] A determination subunit is used to determine the driving state at the next moment corresponding to the target element as the target state, and execute the search for the target element whose difference with the random number is within a preset difference in the element according to the preset random number, until the number of the target states reaches the preset number and the last target state in the target state sequence is the preset value, thereby obtaining the target state sequence.

[0190] Optionally, the splicing module 705 includes:

[0191] An acquisition submodule, used for acquiring a duration ratio of the second travel segment in each first driving condition;

[0192] The splicing submodule is used to splice each of the motion curves according to the duration ratio and the preset duration corresponding to the first driving condition curve to obtain the first driving condition curve.

[0193] Optionally, the driving condition constructing device 700 further includes:

[0194] A training module, used to train the long short-term memory neural network model using the wheel angle data to obtain the recurrent neural network model;

[0195] Optionally, the input module is specifically used to:

[0196] The first driving condition curve and the preset vehicle weight are input into the recurrent neural network model to obtain the second driving condition curve.

[0197] Optionally, the training module includes:

[0198] Inputting the wheel angle data into a long short-term memory neural network model, so that the long short-term memory neural network model performs forward propagation calculation based on a preset first model parameter to obtain a predicted value;

[0199] When a root mean square error between the predicted value and the preset actual value is greater than a target threshold, performing back propagation calculation on the long short-term memory neural network model using the predicted value to obtain a gradient value of the first model parameter;

[0200] Using the gradient value to update the first model parameter to obtain a second model parameter;

[0201] The second model parameter is determined as a new first model parameter, and the step of inputting the wheel angle data into the long short-term memory neural network model so that the long short-term memory neural network model performs forward propagation calculation based on the preset first model parameter to obtain a predicted value, until the root mean square error between the predicted value and the actual value is less than the target threshold, thereby obtaining the recurrent neural network model.

[0202] Optionally, the training module includes:

[0203] Normalizing the wheel angle data to obtain a first data set;

[0204] Converting the first data set into a second data set represented in vector form;

[0205] The second data set is used to train the long short-term memory neural network model to obtain the recurrent neural network model.

[0206] The driving condition construction device 700 provided in the embodiment of the present application can implement each process implemented in the aforementioned method embodiment, and will not be described again here to avoid repetition.

[0207] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0208] Figure 8 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0209] The device may include a processor 801 and a memory 802 storing program instructions.

[0210] When the processor 801 executes the program, the steps in any of the above method embodiments are implemented.

[0211] Exemplarily, the program may be divided into one or more modules / units, one or more modules / units are stored in the memory 802 and executed by the processor 801 to complete the present application. One or more modules / units may be a series of program instruction segments capable of completing a specific function, and the instruction segments are used to describe the execution process of the program in the device.

[0212] Specifically, the processor 801 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0213] The memory 802 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 802 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 802 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 802 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 802 is a non-volatile solid-state memory.

[0214] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) readable storage media (e.g., memory devices) encoded with software including computer executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0215] The processor 801 implements any one of the methods in the above embodiments by reading and executing program instructions stored in the memory 802 .

[0216] In one example, the electronic device may further include a communication interface 803 and a bus 810. The processor 801, the memory 802, and the communication interface 803 are connected via the bus 810 and communicate with each other.

[0217] The communication interface 803 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0218] Bus 810 includes hardware, software or both, and the parts of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 810 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.

[0219] In addition, in combination with the method in the above embodiment, the embodiment of the present application can provide a storage medium for implementation. The storage medium stores program instructions; when the program instructions are executed by a processor, any one of the methods in the above embodiment is implemented.

[0220] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0221] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0222] An embodiment of the present application provides a computer program product, which is stored in a storage medium. The program product is executed by at least one processor to implement the various processes of the above-mentioned method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0223] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0224] The functional modules shown in the above block diagram can be implemented as hardware, software, firmware or their combination. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), suitable firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. Programs or code segments can be stored in machine-readable media, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable media" can include any medium that can store or transmit information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. Code segments can be downloaded via computer grids such as the Internet, intranets, etc.

[0225] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0226] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0227] The above are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A driving condition construction method, characterized in that: The method comprises: In the driving scenario to be tested, obtain the driving information of the vehicle during operation; Dividing the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a driving process of the vehicle between two adjacent parking states; Clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition; For any first driving condition among the plurality of driving conditions, determining a motion curve corresponding to the first driving condition according to a preset vehicle speed interval and vehicle speeds in a plurality of second travel segments in the first driving condition; splicing motion curves corresponding to each first driving condition to obtain a first driving condition curve, wherein the first driving condition curve is used to represent the relationship between the speed and time of the vehicle in the driving scenario to be tested; The first driving condition curve is input into a recurrent neural network model to obtain a second driving condition curve, wherein the second driving condition curve is used to express the vehicle steering angle corresponding to the vehicle at different times and speeds, and the recurrent neural network model is trained using the driving information.

2. The driving condition construction method according to claim 1, characterized in that: Each of the first travel segments includes a plurality of first characteristic parameters, and a parameter value corresponding to each first characteristic parameter; The clustering of the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition comprises: Analyze the multiple first characteristic parameter values ​​included in each first trip segment to obtain key characteristic parameters; Constructing a score matrix corresponding to each of the first trip segments according to the key feature parameters; Clustering is performed based on the score matrix corresponding to each of the first travel segments to obtain a plurality of second travel segments under a variety of driving conditions.

3. The driving condition construction method according to claim 1, characterized in that: For any first driving condition among the multiple driving conditions, determining a motion curve corresponding to the first driving condition according to a preset vehicle speed range and vehicle speeds in a plurality of second travel segments in the first driving condition includes: For any first driving condition among the plurality of driving conditions, establishing a state transition probability matrix corresponding to the first driving condition, wherein the state transition probability matrix is ​​used to represent the probability of the vehicle changing between different driving states under the first driving condition; The state transition probability matrix corresponding to the first driving condition is sampled to determine a motion curve corresponding to the first driving condition.

4. The driving condition construction method according to claim 3, characterized in that: According to the preset vehicle speed interval and the vehicle speeds in a plurality of second travel segments in the first driving condition, a state transition probability matrix corresponding to the first driving condition is established, including: In the vehicle speed interval, determining a target interval corresponding to the vehicle speed in each of a plurality of second travel segments in the first driving condition; According to a preset correspondence between the target interval and the driving state, obtaining the driving state of the vehicle in the plurality of second travel segments; Obtain the driving frequency between driving states at adjacent times; Constructing a state transition frequency matrix corresponding to each driving condition according to the driving frequency; The state transition probability matrix corresponding to each driving condition is established according to the state transition frequency matrix.

5. The driving condition construction method according to claim 3, characterized in that: The sampling of the state transition probability matrix corresponding to the first driving condition to determine the motion curve corresponding to the first driving condition includes: According to a preset first state and a preset random number, searching for a target state sequence in the state transition probability matrix; Converting the target state sequence into an initial speed curve; When a difference between a characteristic parameter error of the initial speed curve and a characteristic parameter error of the first driving condition is within a preset threshold, the initial speed curve is determined as the motion curve.

6. The driving condition construction method according to claim 5, characterized in that: The step of searching for a target state sequence in the state transition probability matrix according to a preset first state and a preset random number includes: Searching for an element whose driving state at the current moment is the first state in the state transition probability matrix, wherein the element represents the probability of the vehicle changing between the driving state at the current moment and the driving state at the next moment; According to a preset random number, searching among the elements for a target element whose difference with the random number is within a preset difference; The driving state at the next moment corresponding to the target element is determined as the target state, and the target element whose difference with the random number is within a preset difference is searched in the element according to the preset random number, until the number of the target states reaches the preset number and the last target state in the target state sequence is the preset value, thereby obtaining the target state sequence.

7. The driving condition construction method according to claim 1, characterized in that: The driving information includes wheel angle data; After the motion curves corresponding to each first driving condition are spliced ​​to obtain the first driving condition curve, the method further includes: The wheel angle data is used to train a long short-term memory neural network model to obtain the recurrent neural network model; The step of inputting the first driving condition curve into a recurrent neural network model to obtain a second driving condition curve comprises: The first driving condition curve and the preset vehicle weight are input into the recurrent neural network model to obtain the second driving condition curve.

8. The driving condition construction method according to claim 7, characterized in that: The method of using the wheel angle data to train the long short-term memory neural network model to obtain the recurrent neural network model includes: Inputting the wheel angle data into a long short-term memory neural network model, so that the long short-term memory neural network model performs forward propagation calculation based on a preset first model parameter to obtain a predicted value; When a root mean square error between the predicted value and the preset actual value is greater than a target threshold, performing back propagation calculation on the long short-term memory neural network model using the predicted value to obtain a gradient value of the first model parameter; Using the gradient value to update the first model parameter to obtain a second model parameter; The second model parameter is determined as a new first model parameter, and the step of inputting the wheel angle data into the long short-term memory neural network model so that the long short-term memory neural network model performs forward propagation calculation based on the preset first model parameter to obtain a predicted value, until the root mean square error between the predicted value and the actual value is less than the target threshold, thereby obtaining the recurrent neural network model.

9. The driving condition construction method according to claim 7, characterized in that: The method of using the wheel angle data to train the long short-term memory neural network model to obtain the recurrent neural network model includes: Normalizing the wheel angle data to obtain a first data set; Converting the first data set into a second data set represented in vector form; The second data set is used to train the long short-term memory neural network model to obtain the recurrent neural network model.

10. A driving condition construction device, characterized in that: include: An acquisition module, used to acquire driving information of the vehicle during operation in the driving scenario to be tested; a dividing module, configured to divide the running process into a plurality of first travel segments according to the driving information, wherein the first travel segment is a running process of the vehicle between two adjacent parking states; A clustering module, used for clustering the plurality of first travel segments to obtain a plurality of driving conditions and a plurality of second travel segments included in each driving condition; An establishing module is used to determine, for any first driving condition among the plurality of driving conditions, a motion curve corresponding to the first driving condition according to a preset vehicle speed interval and vehicle speeds in a plurality of second travel segments in the first driving condition; The splicing module is used to splice the motion curves corresponding to each first driving condition to obtain a first driving condition curve, wherein the first driving condition curve is used to represent the relationship between the speed and time of the vehicle during operation.