CN113689594A - Novel automobile driving condition construction method - Google Patents
Novel automobile driving condition construction method Download PDFInfo
- Publication number
- CN113689594A CN113689594A CN202110917547.8A CN202110917547A CN113689594A CN 113689594 A CN113689594 A CN 113689594A CN 202110917547 A CN202110917547 A CN 202110917547A CN 113689594 A CN113689594 A CN 113689594A
- Authority
- CN
- China
- Prior art keywords
- data
- acceleration
- speed
- kinematic
- automobile
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010276 construction Methods 0.000 title description 4
- 230000001133 acceleration Effects 0.000 claims abstract description 57
- 238000000034 method Methods 0.000 claims abstract description 46
- 238000000513 principal component analysis Methods 0.000 claims abstract description 14
- 238000007621 cluster analysis Methods 0.000 claims abstract description 12
- 239000012634 fragment Substances 0.000 claims description 32
- 238000012545 processing Methods 0.000 claims description 27
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 14
- 230000002159 abnormal effect Effects 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000000694 effects Effects 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 abstract description 10
- 238000000605 extraction Methods 0.000 abstract description 9
- 238000005265 energy consumption Methods 0.000 abstract description 4
- 238000007689 inspection Methods 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000013480 data collection Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C5/00—Registering or indicating the working of vehicles
- G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
- G07C5/0841—Registering performance data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a new method for constructing the running condition of an automobile, because the original collected data directly recorded by the collecting equipment of the running data of the automobile often contains some bad data, a reasonable method is needed to be designed to preprocess the original bad data, thereby finishing the extraction of the kinematic segment, according to the preprocessed data, a curve of the running condition of the automobile which can embody the running characteristics of the automobile participating in the data collection is generated by utilizing a principal component analysis and a cluster analysis method, and compared with the original data, and the inspection result is summarized. Selecting and calculating 9 characteristic parameters, trying a new division standard of acceleration, deceleration, idling and constant speed working conditions, and obtaining a driving working condition based on the actual speed, acceleration and other characteristic parameters of the urban typical road; the obtained automobile driving condition curve is beneficial to relevant departments to formulate a test condition reflecting the actual road driving condition, and further the management level of automobile energy consumption is improved.
Description
Technical Field
The invention relates to a data driving method, in particular to a novel method for constructing a running condition of an automobile.
Background
The Driving Cycle (also called vehicle test Cycle) is a speed-time curve describing the Driving of the vehicle. The driving condition is used as the basis of a vehicle energy consumption/emission test method and a limit value standard, reflects the kinematic characteristics of the road driving of the automobile, is an important and common basic technology in the automobile industry, and is also a main reference [1] for calibrating and optimizing various performance indexes of the automobile. In the beginning of this century, China directly adopts the certification of new European Standard running cycle (NEDC) running conditions for energy consumption/emission of automobile products [2 ]. In recent years, along with the rapid increase of automobile reserves, road traffic conditions in China are greatly changed, and governments, enterprises and people gradually find that the deviation between the actual oil consumption and the result of regulation certification is larger and larger by taking the NEDC working condition as the reference for optimizing and calibrating the automobiles [3 ]. In europe, many defects of NEDC operation have been found in many years of practice, and the world light vehicle test cycle (WLTC) is adopted, but the two most important characteristics of idle time ratio and average speed of the operation are different from the actual automobile running operation in China [4 ]. On the other hand, the territory of China is wide, and the development degree, climatic conditions and traffic conditions of each city are different, so that the driving condition characteristics of the automobiles in each city are obviously different [5 ]. In order to better understand the importance of establishing the automobile driving condition curve, the method is deeply researched to serve as the basis of vehicle development and evaluation, and the test condition reflecting the actual road driving condition in China is formulated, so that the method is more and more important.
Disclosure of Invention
Aiming at the problems in the prior art, because the original collected data directly recorded by the collecting equipment of the automobile running data often contains some bad data, a reasonable method is needed to be designed to preprocess the original bad data, so that the extraction of the kinematic segment is completed, an automobile running condition curve which can reflect the running characteristics of the automobile participating in data collection is generated according to the preprocessed data by using a principal component analysis and cluster analysis method, and the curve is compared with the original data to summarize the inspection result.
In order to achieve the purpose, the invention adopts the technical scheme that: a new method for constructing the running condition of an automobile comprises the following steps:
step one, data preprocessing:
the method comprises the following specific steps of carrying out analysis pretreatment on the condition of influence on data:
(1) processing missing data of a GPS signal, wherein discontinuous segment data is changed into continuous segment data; analyzing through a time sequence to judge the continuity of the time data; directly extracting continuous time segments, and thus directly performing data analysis on each continuous time segment;
(2) aiming at data processing of long-time low-speed driving, directly assigning the speed of the data to be 0m/s, wherein the data at the moment is still a continuous time period;
(3) processing the abnormal conditions of acceleration and deceleration, directly removing abnormal data, and changing the time data of a continuous section into the time data of a discontinuous section; calculating the acceleration of the vehicle at each moment according to the acceleration formula at each moment, wherein the formula (1) is as follows:
in the formula, ak,k-1Acceleration between the kth second and the k-1 th second, vkIs the kth second speed, tkAt the kth second moment, N is the total number of points in the continuous section; since it is considered that the acceleration time of a car from 0km/h to 100km/h is more than 7s, the maximum acceleration of a general car is 3.968m/s2While the maximum deceleration is 7.5 to 8m/s2According to the two judgment standards, eliminating the data which are not in the acceleration range;
(4) because the data with abnormal acceleration and deceleration is removed in the step (3), part of the data in the continuous time section becomes discontinuous data; the processing method of the step (1) needs to be continuously repeated on the discontinuous data segments to enable the discontinuous data segments to become continuous data segments;
(5) for the processing of idle speed data, deleting the data part with the idle speed time exceeding 180s for each continuous time period, and reserving the idle speed part less than 180 s; since the processing of the data takes place in successive segments, successive time segments can still be derived in the end.
Step two, extracting the kinematic fragment
Extracting a plurality of kinematic segments from the effective data processed in the step one;
the kinematic segment refers to a vehicle speed interval between the start of an idle state and the start of the next idle state of the automobile, and the driving process of the automobile is regarded as the combination of a plurality of kinematic segments;
there are two main methods of kinematic segmentation, one starting at the idle start and ending at the next idle start; the other starts at the idle speed end and ends at the next idle speed end;
step three, establishing the running condition of the automobile:
(1) selecting characteristic parameters: determining 9 characteristic parameters to represent the characteristic parameters and the kinematic fragment information through a kinematic fragment database obtained after data processing, and calculating the characteristic parameters of each kinematic fragment to obtain a kinematic fragment characteristic parameter database;
the 9 characteristic parameters are respectively the average speed vmAverage traveling speed vrAverage acceleration awAverage deceleration adIdle time ratio P0Acceleration time ratio PwSpeed reduction time ratio PdStandard deviation of velocity σvAcceleration standard deviation sigmaa；
(2) And (3) main component analysis: by using principal component analysis, the number of data variables is reduced as much as possible under the condition of not losing too much information of original variables, data dimensionality is reduced, calculation is simplified, and 3 principal component variables are obtained to replace original 9 characteristic parameter data, namely, 3 principal component variables are obtained by calculating the data after the covariance matrix, the characteristic value and the original variables are standardized through SPSS software;
(3) K-Means Cluster analysis:
the core of the cluster analysis algorithm is to determine a central point, and for a given sample set, the sample set is divided into K clusters according to the distance between samples; the selected K value determines the clustering effect of the samples; let the distance of the point in the cluster as little as possible, and let the distance between the cluster as big as possible, the clustering distance that adopts is euclidean distance, as shown in equation (14):
performing K-Means cluster analysis on the 3 principal components obtained by analyzing the principal components in the step three (2) by using a procfastplus process in SAS software; the data analyzed by the principal components are gathered into three types of running working conditions, wherein the three types of running working conditions are respectively congestion type 1, smooth type 2 and special congestion type 3;
(4) curve fitting of operating conditions
The closer the distance from the final convergence point to the kinematic segment of the automobile driving condition curve, the more representative the overall automobile driving condition curve of the category;
and extracting a certain proportion of fragments from each clustered category for fitting, and finally combining the fragments of each category to form an automobile driving condition curve with the length within a specified time.
Further, the kinematic segment analysis is performed on the first scheme in the second step, so that an idling starting point of an idling period and a point at which the last speed of the kinematic segment is 0m/s need to be found out, and the kinematic segment is divided;
the idling starting point of the idling time period, namely the speed and the acceleration are 0; the same point as the last point with the speed of 0m/s of the kinematic segment; because in a complete kinematic segment, more than one point with the same speed and acceleration characteristics of the idle starting point is provided, and only one point satisfying the idle end point characteristics is provided; therefore, only the point with the last speed of 0m/s of the continuous segment needs to be found; it is known that the last point at a speed of 0m/s simultaneously satisfies the following characteristic as shown in formula (2):
furthermore, in the third step, (2) principal component analysis applies principal component analysis, the number of data variables is reduced as much as possible under the condition that too much information of original variables is not lost, data dimensionality is reduced, and calculation is simplified; the method for calculating the contribution ratio of each principal component is shown in formula (13):
wherein λ is1，λ2，…，λpCharacteristic values corresponding to the 1 st principal component to the p-th principal component;
performing principal component analysis on 9 parameters of all the kinematic segments by using SPSS software, calculating to obtain the contribution rate and the accumulated contribution rate of each principal component, analyzing the accumulated contribution rate of the first 3 principal components, and verifying that it is reasonable that the principal components replace the parameters; therefore, 3 principal component variables are selected to replace original 9 characteristic parameter data, and the 3 principal component variables are calculated through the covariance matrix, the characteristic value and the data after the original variables are standardized, which are obtained through SPSS software.
Further, in the third step (3), the classified overall characteristic parameter characteristic values of the various classes are obtained through calculation by means of Excel and SPSS, so that the condition that the class 1 represents the urban road congestion condition, the class 2 kinematic segment represents the urban road smoothness condition, and the class 3 represents the urban road special congestion condition is obtained.
The invention has the beneficial effects that: the invention discloses a new method for constructing the running condition of an automobile, which provides a new idea for establishing the running condition of the actual road in China; in the preprocessing process, a huge data set is processed by means of Matlab, the data are preprocessed by a reasonable method to remove bad data, the data are subjected to kinematic fragment extraction, and characteristic parameters of the kinematic fragments are calculated; and performing dimensionality reduction on the characteristic parameters by adopting principal component analysis, and clustering the principal component analyzed data into three types of running conditions, namely smooth running, relatively congested running and congested running, on the basis of clustering analysis. Selecting kinematic segments from the three types of running conditions respectively, constructing a running condition curve, selecting and calculating 9 characteristic parameters, trying a new division standard of acceleration, deceleration, idling and uniform speed working conditions, and obtaining a running condition based on the characteristic parameters of actual speed, acceleration and the like of the typical urban road; the obtained automobile driving condition curve is beneficial to relevant departments to formulate a test condition reflecting the actual road driving condition, and further the management level of automobile energy consumption is improved.
Drawings
FIG. 1 is a data processing process diagram;
FIG. 2 is an explanatory diagram of a kinematic fragment;
FIG. 3 is a fast clustering scatter plot;
FIG. 4 is a fitting curve diagram of the driving condition of the automobile;
FIG. 5 is a comparison of a fitted curve and an actual curve of driving conditions.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood, however, that the description herein of specific embodiments is only intended to illustrate the invention and not to limit the scope of the invention.
According to more than 40 ten thousand road running records collected by the same vehicle in different time periods, the aim is to establish a vehicle running condition curve which can reflect the running characteristics of the vehicle participating in data collection. The motion characteristics (such as average speed, average acceleration and the like) of the automobile represented by the curve can represent corresponding characteristics of the collected data source, and the smaller the error between the motion characteristics and the corresponding characteristics, the better the representativeness of the constructed automobile running condition is.
Because the original collected data directly recorded by the collecting equipment of the automobile driving data often contain some bad data, a reasonable method needs to be designed to preprocess the original bad data, so as to complete the extraction of the kinematic segment. According to the preprocessed data, a novel automobile driving condition construction method is provided, a main component analysis and cluster analysis method is utilized to generate an automobile driving condition curve which can reflect the driving characteristics of the automobile participating in data acquisition, the curve is compared with the original data, and the inspection result is summarized; the specific contents are as follows:
1 vehicle driving condition data preprocessing
1.1 data error analysis method
Because the original data that collection equipment of car driving data directly recorded often can contain some bad data values, can cause certain error to the data test data of gathering, include because reasons such as high-rise building cover lead to GPS signal loss, the car acceleration data is unusual and long-time idle speed appears. The reasons and processes for error generation are as follows:
(1) time discontinuity
The time discontinuity is considered to be caused by the fact that the GPS positioning is unsuccessful due to high-rise building coverage or tunnel crossing and the like, so that the time discontinuity in the original data is caused. For such errors, successive time slices can be directly extracted, so that data analysis is directly performed on each successive time slice.
(2) Acceleration and deceleration anomaly
Generally, the acceleration time of the automobile within 0km/h to 100km/h is more than 7s, and the maximum deceleration of the emergency brake is 7.5m/s2To 8m/s2. The abnormal acceleration and deceleration values can be screened out by calculating the acceleration and deceleration of each time period, and then are directly removed, so that the error generated in the data acquisition process is reduced.
(3) Long idle speed
The reason for idling the automobile for a long time is two, the first reason is that the automobile is stopped for a long time (for example, the automobile is not stopped for waiting for people, the automobile is stopped for a long time, but the acquisition equipment is still running, and the like); the second long-time traffic jam, intermittent low-speed driving, and the like cause a long-time idling. The longest idling time of the long time is processed according to 180s, for the reasons of the two idling, the idling time does not exceed 180s, the original data are reserved, and the data of the idling section exceeding 180s are regarded as abnormal values to be removed, so that the influence of the data on the later result is reduced.
According to the error analysis of the data, the data analysis and processing are sequentially performed on the conditions possibly influencing the data, and the original data is processed mainly through 5 steps, which specifically include the following steps:
(6) for the processing of missing data of the GPS signal, discontinuous segments of data may be changed into continuous segments of data at this time. To determine the continuity of the time data, analysis was performed by time series. And directly extracting the continuous time segments, thereby directly carrying out data analysis on each continuous time segment.
(7) For data processing of long-time low-speed driving, the speed of the data is directly assigned to be 0m/s, and the data at the moment is still continuous time periods.
(8) The processing of the addition and deceleration abnormal conditions directly removes the abnormal data, and the time data of the continuous section becomes the time data of the discontinuous section. Calculating the acceleration of the vehicle at each moment according to the acceleration formula at each moment, wherein the formula (1) is as follows:
in the formula, ak,k-1Acceleration between the kth second and the k-1 th second, vkIs the kth second speed, tkAt the time of the k second, N is the total number of points in the continuous section. Since it is considered that the acceleration time of a car from 0km/h to 100km/h is more than 7s, the maximum acceleration of a general car is 3.968m/s2While the maximum deceleration is 7.5 to 8m/s2And according to the two judgment standards, eliminating the data which is not in the acceleration range.
(9) Because the data with abnormal acceleration and deceleration is removed in the step (3), part of the continuous time section data becomes discontinuous data. The processing method of step (1) is continuously repeated for the discontinuous data segments to become continuous data segments.
(10) For the processing of idle data, data parts with idle time exceeding 180s are deleted for each continuous time period, and idle parts smaller than 180s are reserved. Since the processing of the data takes place in successive segments, successive time segments can still be derived in the end.
In order to more clearly and intuitively show the data processing process, fig. 1 selects original data of a certain continuous segment, and a three-dimensional graph representing the change of the data segment in each data preprocessing step (1) to (5) is drawn through Matlab.
1.2 data preprocessing results
And processing the collected original data of the three files in different time periods by utilizing Matlab according to the steps, and outputting a new Excel file. Wherein, the file 1 is the collected data of 12 months and 3 weeks, the file 2 is the collected data of 11 months and 1 week, and the file 3 is the collected data of 12 months and 1 week. The pretreatment results are shown in table 1: the number of data point records left after the file 1 is subjected to data processing is 178771, the number of data point records left after the file 2 is subjected to data processing is 139106, and the number of data point records left after the file 3 is subjected to data processing is 158198, which is 476075 records. A plurality of kinematic segments are needed for constructing the automobile driving condition curve, and then a plurality of kinematic segments are extracted from the processed effective data.
TABLE 1 number of remaining data points recorded
2 kinematic fragment extraction
2.1 principle of extraction
From a starting point to a destination, the automobile is influenced by the road type and traffic conditions and passes through a series of processes of starting, low speed, stopping and the like for multiple times. The kinematic segment refers to the speed interval of the vehicle from the start of the idle state to the start of the next idle state. Fig. 2 is an explanatory diagram defining a kinematic segment, and the whole kinematic segment includes idle speed, acceleration, deceleration, uniform speed and the like. The driving process of a motor vehicle is considered herein as a combination of a plurality of kinematic segments.
The method comprises the steps of constructing a model of the running working condition of an automobile, considering the whole process as continuous by a classical construction strategy, analyzing the whole process by using statistics, dividing the working condition according to different road region types, constructing a plurality of candidate working conditions for the running of the automobile, and finally selecting all levels of candidate working conditions to be called as a total working condition. Although this method meets the driving condition requirements of the structure, the processing amount is too large, and the correlation between two points can be hidden, so that the data information is repeated. Not only does this increase the computational burden, but also the summary of data information is induced and biased by taking overlapping invalid data information into account when constructing the condition. Therefore, on the basis of data preprocessing, a more superior kinematic fragment method is adopted to analyze the data.
2.2 extraction method
There are two main methods of kinematic segmentation, one starting at the idle start and ending at the next idle start; the other starts at the idle end and ends at the next idle end. The first solution is provided for the analysis of the kinematic segment, so that the starting point of idle speed in the idle period and the last point of 0m/s speed in the kinematic segment are found to divide the kinematic segment.
Each whole piece of kinematic segment has been found in data pre-processing, and the abnormal data is culled and transformed, but not an independent kinematic segment. Table 2 summarizes the velocity and acceleration characteristics for different motion states (see fig. 2 for different motion states) in a kinematic segment.
TABLE 2 speed and acceleration characteristics of each motion state
Through analysis, the idling starting point (the speed and the acceleration take the same value as 0) of the idling time period and the last point of 0m/s of the speed of the kinematic segment are the same point. Since there is more than one point in a complete kinematic segment that is identical to the speed and acceleration characteristics of the idle start point, there is only one point that satisfies the idle end point characteristics. So we only need to find the last point of the consecutive segment with a velocity of 0 m/s. It is known that the last point at a speed of 0m/s simultaneously satisfies the following characteristic as shown in formula (2):
2.3 extraction results
And combining the characteristics of the idle speed point, and extracting the kinematic segment by using a Matlab loop statement. After the Matlab software is subjected to traversal and cyclic extraction, the number of the kinematic fragments of the file 1 is 840, the number of the kinematic fragments of the file 2 is 502, the number of the kinematic fragments of the file 3 is 560, and 1902 kinematic fragments are obtained in total.
3 establishing the running condition of the automobile
3.1 selection of characteristic parameters
It is necessary to analyze the rationality of the driving condition of the constructed vehicle by using certain characteristic parameters. A reasonable automobile motion characteristic evaluation system needs to consider a plurality of factors, wherein the speed and the acceleration are most important, and the divided kinematics segments cannot be completely described only by the two factors, so that other characteristic parameters still need to be introduced to describe the kinematics segments, too many parameters consume a large amount of calculation time and increase the calculation difficulty, and too few parameters cannot accurately represent a large amount of data[7]. The method selects 9 characteristic parameters which are important to the running condition of the automobile and respectively refer to the average speed vmAverage traveling speed vrAverage acceleration awAverage deceleration adIdle time ratio P0Acceleration time ratio PwSpeed reduction time ratio PdStandard deviation of velocity σvAcceleration standard deviation sigmaa。
The calculation formula of the characteristic parameters is as follows:
(1) average velocity vm
In the formula, TtotalFor each kinematic segmentThe total duration.
(2) Average running speed vr
In the formula, vk +Is v isk>The speed of the motor at 0 is,
the non-idle time of the automobile.
(3) Average acceleration aw
In the formula, ak +The value of the automobile driving acceleration is larger than 0,
the total acceleration duration of the vehicle.
(4) Average deceleration ad
In the formula, ak -The value of the automobile driving acceleration is less than 0,
the total deceleration duration of the vehicle.
(5) Idle time ratio P0
In the formula, T0Is the total idle time.
(6) Acceleration time ratio Pw
(7) Deceleration time ratio Pd
(8) Standard deviation of speed sigmav
(9) Acceleration standard deviation sigmaa
In the formula, amThe calculation formula is shown as follows:
3.2 principal Components analysis
And determining 9 characteristic parameters to represent the kinematic fragment information through a kinematic fragment database obtained after data processing, and calculating the characteristic parameters of each kinematic fragment to obtain the kinematic fragment characteristic parameter database. Because the data volume of the characteristic parameters of the kinematic segment is large, if the calculation volume generated by directly analyzing the 9 characteristic parameters is large, the data dimension reduction processing needs to be carried out through principal component analysis[8]. By using principal component analysis, the number of data variables is reduced as much as possible under the condition of not losing too much information of original variables, the data dimension is reduced, and the calculation is simplified[9-10]。
The method for calculating the contribution ratio of each principal component is shown in formula (13):
wherein λ is1，λ2，…，λpThe characteristic values corresponding to the 1 st principal component to the p-th principal component.
The SPSS software is used to perform principal component analysis on the 1902 kinematic segments of 9 parameters, and the contribution rate and the cumulative contribution rate of each principal component can be calculated, and from Table 3, it can be found that the cumulative contribution rate of the first 3 principal components is 83.690%, which indicates that it is reasonable to replace these parameters with principal components.
TABLE 3 contribution rate of each principal component and cumulative contribution rate
Therefore, 3 principal component variables can be selected to replace original 9 characteristic parameter data, and the data obtained by normalizing the covariance matrix, the eigenvalue and the original variables through SPSS software can be calculated to obtain 3 principal component variables so as to carry out next clustering analysis.
3.3K-Means Cluster analysis
The core of the cluster analysis algorithm is to determine a central point, and for a given sample set, divide the sample set into K clusters according to the distance between samples. The selected K value determines the clustering effect of the samples. The distance between the points in the clusters is made as small as possible, and the distance between the clusters is made as large as possible[11]. The clustering distance used herein is the Euclidean distance, as shown in equation (14):
and (3) performing K-Means clustering analysis on the three main components by using a procfastplus process in SAS software on the 3 main components obtained by analyzing the main components in the 3.2. The effect of classifying into 3 classes is better by multiple clustering analysis. Plot scatter plot 3 is as follows:
wherein the first class of fragments comprises 805 fragments, the second class comprises 1088 fragments, and the third class comprises 9 fragments. Table 4 intercepts the clustering results of part of the kinematic segments of the clustering results, and all the distance results of the kinematic segments are only shown in part of the following table:
TABLE 4 kinematic fragment clustering results
Calculating by using Excel and SPSS to obtain the characteristic value of the classified overall characteristic parameter of each category, as shown in the following table 5:
TABLE 5 characteristic parameters of three types of kinematic segments
As can be seen from Table 5, the average speed of the kinematics segment class 2 is 22.64km/h, which is much higher than class 1 and class 3, while the idle ratio is 0.19, which is lower than the other two classes. It can be seen that the class 2 kinematic segment represents an urban road clear situation. The average acceleration and deceleration of class 1 is much lower than class 3. We can see that category 1 represents urban road congestion, while category 3 represents urban road congestion in particular.
3.4 working condition curve fitting
On the basis of the cluster analysis, whether each kinematic segment can represent the kinematic segment of the category can be examined through the distance between the kinematic segment and the final cluster point[12]. The closer the kinematic segment is to the final convergence point, the more representative the vehicle driving curve of the segment is of the overall vehicle driving curve of the class[13]。
And extracting segments with a certain proportion from each clustered class for fitting, and finally combining the segments of each class to form an automobile driving condition curve with the length of 1200-1300 s. Since the number of the kinematic segments of the 3 rd category is small, for the convenience of calculation, we only select the first 5 kinematic segments in the 3 rd category closest to the final rendezvous point. And (3) the kinematic segments of the 1 st class and the 2 nd class, and the first 20 kinematic segments which are closest to the final focus point are selected respectively. By interpolation, the 1 st kinematic segment of the 3 rd class, the 5 st kinematic segment of the 1 st class, and the 7 nd kinematic segment of the 2 nd class are selected, and 13 kinematic segments are counted.
Matlab software is used for combining the kinematic segments to construct a 1300 s-long automobile driving condition fitting curve as shown in FIG. 4, wherein the kinematic segment in the [0s,9s ] interval is a 3 rd type kinematic segment, the kinematic segment in the [10s,458s ] interval is a 1 st type kinematic segment, and the kinematic segment in the [459s,1300s ] interval is a 2 nd type kinematic segment.
3.5 rationality test
3.5.1 characteristic parameter error analysis test
The operating curve is constructed to represent the corresponding characteristics of the actual collected data source (processed data) according to requirements. The smaller the error between the constructed working condition curve and the actually acquired data curve is, the better the representativeness of the constructed automobile running working condition is. In order to more intuitively display the error between the working condition curve constructed in the graph 4 and the actually acquired data curve, 13 kinematic segments are randomly extracted from 1902 kinematic segments through an unidrnd function in Matlab to draw a random working condition curve. As shown in FIG. 5, the error between the driving condition curve and the actually acquired data curve is small, and the fitting effect is good.
The basic idea of characteristic parameter error analysis is to compare a running condition fitting curve with a random actually acquired data curve, respectively calculate the relative error of the constructed road running condition and the whole corresponding to the road running condition from 9 characteristic parameters defined by 3.1 as analysis indexes, and analyze whether the constructed working condition is effective or not according to the calculated result. In general, a relative error of less than 10% in the study is within an acceptable range[14]. Relative to each otherThe error calculation formula is shown in equation (15):
δifor a characteristic parameter, delta, of the constructed road driving conditionsjThe category corresponds to the overall characteristic parameter. The calculation results are shown in table 6:
TABLE 6 running condition error analysis
From table 6, we can find that the relative errors of the average speed, the average driving speed and the idle time ratio of each category are within 7.04%, so the fitting effect of the part of data is reasonable. The errors of the characteristic parameters of the 2 nd category are all within 7.50%. The acceleration time ratio in the category 3 is 61.63%, the error is large, but the constructed automobile running condition is reasonable considering that only the extreme case of the data with large relative error is in an acceptable range.
3.5.2 correlation coefficient test
The correlation coefficient is a variable describing the correlation between random variables X and Y[15]The parameter characteristic value sequences of each segment of each class are regarded as the distribution of random variables X, and the parameter characteristic value sequences of various types of synthesis are regarded as the distribution of random variables Y. The Pearson correlation coefficient calculation formula is shown in equation (16):
the correlation coefficient of X and Y is calculated through the correlation of SPSS bivariate, the closer the number is to 1, the more linear correlation between the parameter characteristic value of the segment and the parameter characteristic value of the whole segment is, the more representative the data of the segment is[16]. Table 7 lists the correlation coefficient comparisons corresponding to the selected segments in the three classes, i.e., eachAnd (4) checking correlation coefficients of the representative characteristic parameters of the similar road running conditions and the overall parameters of the road running conditions.
Table 7 correlation coefficient test results
In table 7, we can obtain three types of correlation coefficients all close to 1, which indicates that the linear correlation between the parameter characteristic value of the segment and the parameter characteristic value of the whole segment is large. Therefore, the data of the selected segments are more representative, and the inspection result is reasonable.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents or improvements made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (5)
1. A new method for constructing the running condition of an automobile is characterized in that: the method comprises the following steps:
step one, data preprocessing:
the method comprises the following specific steps of carrying out analysis pretreatment on the condition of influence on data:
(1) processing missing data of a GPS signal, wherein discontinuous segment data is changed into continuous segment data; analyzing through a time sequence to judge the continuity of the time data; directly extracting continuous time segments, and thus directly performing data analysis on each continuous time segment;
(2) aiming at data processing of long-time low-speed driving, directly assigning the speed of the data to be 0m/s, wherein the data at the moment is still a continuous time period;
(3) processing the abnormal conditions of acceleration and deceleration, directly removing abnormal data, and changing the time data of a continuous section into the time data of a discontinuous section; calculating the acceleration of the vehicle at each moment according to the acceleration formula at each moment, wherein the formula (1) is as follows:
in the formula, ak,k-1Acceleration between the kth second and the k-1 th second, vkIs the kth second speed, tkAt the kth second moment, N is the total number of points in the continuous section; since it is considered that the acceleration time of a car from 0km/h to 100km/h is more than 7s, the maximum acceleration of a general car is 3.968m/s2While the maximum deceleration is 7.5 to 8m/s2According to the two judgment standards, eliminating the data which are not in the acceleration range;
(4) because the data with abnormal acceleration and deceleration is removed in the step (3), part of the data in the continuous time section becomes discontinuous data; the processing method of the step (1) needs to be continuously repeated on the discontinuous data segments to enable the discontinuous data segments to become continuous data segments;
(5) for the processing of idle speed data, deleting the data part with the idle speed time exceeding 180s for each continuous time period, and reserving the idle speed part less than 180 s; since the processing of the data takes place in successive segments, successive time segments can still be derived in the end.
Step two, extracting the kinematic fragment
Extracting a plurality of kinematic segments from the effective data processed in the step one;
the kinematic segment refers to a vehicle speed interval between the start of an idle state and the start of the next idle state of the automobile, and the driving process of the automobile is regarded as the combination of a plurality of kinematic segments;
there are two main methods of kinematic segmentation, one starting at the idle start and ending at the next idle start; the other starts at the idle speed end and ends at the next idle speed end;
step three, establishing the running condition of the automobile:
(1) selecting characteristic parameters: determining 9 characteristic parameters to represent the characteristic parameters and the kinematic fragment information through a kinematic fragment database obtained after data processing, and calculating the characteristic parameters of each kinematic fragment to obtain a kinematic fragment characteristic parameter database;
the 9 characteristic parameters are respectively the average speed vmAverage traveling speed vrAverage acceleration awAverage deceleration adIdle time ratio P0Acceleration time ratio PwSpeed reduction time ratio PdStandard deviation of velocity σvAcceleration standard deviation sigmaa；
(2) And (3) main component analysis: by using principal component analysis, the number of data variables is reduced as much as possible under the condition of not losing too much information of original variables, data dimensionality is reduced, calculation is simplified, and 3 principal component variables are obtained to replace original 9 characteristic parameter data, namely, 3 principal component variables are obtained by calculating the data after the covariance matrix, the characteristic value and the original variables are standardized through SPSS software;
(3) K-Means Cluster analysis:
the core of the cluster analysis algorithm is to determine a central point, and for a given sample set, the sample set is divided into K clusters according to the distance between samples; the selected K value determines the clustering effect of the samples; let the distance of the point in the cluster as little as possible, and let the distance between the cluster as big as possible, the clustering distance that adopts is euclidean distance, as shown in equation (14):
performing K-Means cluster analysis on the 3 principal components obtained by analyzing the principal components in the step three (2) by using a procfastplus process in SAS software; the data analyzed by the principal components are gathered into three types of running working conditions, wherein the three types of running working conditions are respectively congestion type 1, smooth type 2 and special congestion type 3;
(4) curve fitting of operating conditions
The closer the distance from the final convergence point to the kinematic segment of the automobile driving condition curve, the more representative the overall automobile driving condition curve of the category;
and extracting a certain proportion of fragments from each clustered category for fitting, and finally combining the fragments of each category to form an automobile driving condition curve with the length within a specified time.
2. The method for constructing the driving condition of the automobile as claimed in claim 1, wherein: performing the kinematic segment analysis on the first scheme in the step two, so that an idling starting point of an idling time period and a point at which the last speed of the kinematic segment is 0m/s need to be found out, and dividing the kinematic segment;
the idling starting point of the idling time period, namely the speed and the acceleration are 0; the same point as the last point with the speed of 0m/s of the kinematic segment; because in a complete kinematic segment, more than one point with the same speed and acceleration characteristics of the idle starting point is provided, and only one point satisfying the idle end point characteristics is provided; therefore, only the point with the last speed of 0m/s of the continuous segment needs to be found; it is known that the last point at a speed of 0m/s simultaneously satisfies the following characteristic as shown in formula (2):
3. the method for constructing the driving condition of the automobile as claimed in claim 1, wherein: the calculation formula of the selected characteristic parameters of the characteristic parameters in step three (1) is as follows:
(1) average velocity vm
In the formula, TtotalFor each kinematic segment the total duration.
(2) Average running speed vr
In the formula, vk +Is v isk>The speed of the motor at 0 is,
the non-idle time of the automobile.
(3) Average acceleration aw
In the formula, ak +The value of the automobile driving acceleration is larger than 0,
the total acceleration duration of the vehicle.
(4) Average deceleration ad
In the formula, ak -The value of the automobile driving acceleration is less than 0,
the total deceleration duration of the vehicle.
(5) Idle time ratio P0
In the formula, T0Is the total idle time.
(6) Acceleration time ratio Pw
(7) Speed reductionTime ratio Pd
(8) Standard deviation of speed sigmav
(9) Acceleration standard deviation sigmaa
In the formula, amThe calculation formula is shown as follows:
4. the method for constructing the driving condition of the automobile as claimed in claim 1, wherein: in the third step, (2) principal component analysis applies principal component analysis, the number of data variables is reduced as much as possible under the condition of not losing too much information of original variables, data dimensionality is reduced, and calculation is simplified; the method for calculating the contribution ratio of each principal component is shown in formula (13):
wherein λ is1，λ2，…，λpCharacteristic values corresponding to the 1 st principal component to the p-th principal component;
performing principal component analysis on 9 parameters of all the kinematic segments by using SPSS software, calculating to obtain the contribution rate and the accumulated contribution rate of each principal component, analyzing the accumulated contribution rate of the first 3 principal components, and verifying that it is reasonable that the principal components replace the parameters; therefore, 3 principal component variables are selected to replace original 9 characteristic parameter data, and the 3 principal component variables are calculated through the covariance matrix, the characteristic value and the data after the original variables are standardized, which are obtained through SPSS software.
5. The method for constructing the driving condition of the automobile as claimed in claim 1, wherein: in the third step (3), the classified overall characteristic parameter characteristic values of all classes are obtained through calculation by means of Excel and SPSS, the condition that the class 1 represents the urban road congestion condition, the class 2 kinematic segment represents the urban road smoothness condition, and the class 3 represents the urban road special congestion condition is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110917547.8A CN113689594A (en) | 2021-08-10 | 2021-08-10 | Novel automobile driving condition construction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110917547.8A CN113689594A (en) | 2021-08-10 | 2021-08-10 | Novel automobile driving condition construction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113689594A true CN113689594A (en) | 2021-11-23 |
Family
ID=78579335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110917547.8A Pending CN113689594A (en) | 2021-08-10 | 2021-08-10 | Novel automobile driving condition construction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113689594A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114136656A (en) * | 2021-11-29 | 2022-03-04 | 中国第一汽车股份有限公司 | Method for constructing motor bench test working condition of pure electric operation automobile |
| CN114444280A (en) * | 2022-01-15 | 2022-05-06 | 重庆长安汽车股份有限公司 | Method for testing fuel consumption of vehicle under high-speed working condition |
| CN115188205A (en) * | 2022-07-04 | 2022-10-14 | 武汉理工大学 | Automobile driving condition correction method based on road information |
| CN115447554A (en) * | 2022-10-28 | 2022-12-09 | 北京易控智驾科技有限公司 | Method and device for detecting braking performance of unmanned mining vehicle, electronic equipment and storage medium |
| CN116776229A (en) * | 2023-08-17 | 2023-09-19 | 深圳市城市交通规划设计研究中心股份有限公司 | Method for dividing typical running conditions of automobile for carbon factor emission |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170147930A1 (en) * | 2015-01-30 | 2017-05-25 | Hewlett Packard Enterprise Development Lp | Performance testing based on variable length segmentation and clustering of time series data |
| CN107329429A (en) * | 2017-07-13 | 2017-11-07 | 中国汽车技术研究中心 | A kind of vehicle carried data collecting terminal built for Light-duty Vehicle driving cycle |
| US20180017402A1 (en) * | 2016-07-15 | 2018-01-18 | Tata Consultancy Services Limited | Method and system for vehicle speed profile generation |
| CN110298579A (en) * | 2019-06-26 | 2019-10-01 | 重庆大学 | A kind of construction method of new energy passenger car urban operating condition |
| CN110717147A (en) * | 2019-10-10 | 2020-01-21 | 辽宁工程技术大学 | Method for constructing driving condition of automobile |
| CN111832225A (en) * | 2020-07-07 | 2020-10-27 | 重庆邮电大学 | Method for constructing driving condition of automobile |
| CN112885080A (en) * | 2021-01-11 | 2021-06-01 | 重庆长安新能源汽车科技有限公司 | Construction method for driving condition of new energy automobile |
| CN113052196A (en) * | 2019-12-27 | 2021-06-29 | 南京理工大学 | Automobile working condition construction method based on greedy algorithm |
-
2021
- 2021-08-10 CN CN202110917547.8A patent/CN113689594A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170147930A1 (en) * | 2015-01-30 | 2017-05-25 | Hewlett Packard Enterprise Development Lp | Performance testing based on variable length segmentation and clustering of time series data |
| US20180017402A1 (en) * | 2016-07-15 | 2018-01-18 | Tata Consultancy Services Limited | Method and system for vehicle speed profile generation |
| CN107329429A (en) * | 2017-07-13 | 2017-11-07 | 中国汽车技术研究中心 | A kind of vehicle carried data collecting terminal built for Light-duty Vehicle driving cycle |
| CN110298579A (en) * | 2019-06-26 | 2019-10-01 | 重庆大学 | A kind of construction method of new energy passenger car urban operating condition |
| CN110717147A (en) * | 2019-10-10 | 2020-01-21 | 辽宁工程技术大学 | Method for constructing driving condition of automobile |
| CN113052196A (en) * | 2019-12-27 | 2021-06-29 | 南京理工大学 | Automobile working condition construction method based on greedy algorithm |
| CN111832225A (en) * | 2020-07-07 | 2020-10-27 | 重庆邮电大学 | Method for constructing driving condition of automobile |
| CN112885080A (en) * | 2021-01-11 | 2021-06-01 | 重庆长安新能源汽车科技有限公司 | Construction method for driving condition of new energy automobile |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114136656A (en) * | 2021-11-29 | 2022-03-04 | 中国第一汽车股份有限公司 | Method for constructing motor bench test working condition of pure electric operation automobile |
| CN114136656B (en) * | 2021-11-29 | 2023-12-08 | 中国第一汽车股份有限公司 | Construction method for motor bench test working condition of pure electric commercial vehicle |
| CN114444280A (en) * | 2022-01-15 | 2022-05-06 | 重庆长安汽车股份有限公司 | Method for testing fuel consumption of vehicle under high-speed working condition |
| CN115188205A (en) * | 2022-07-04 | 2022-10-14 | 武汉理工大学 | Automobile driving condition correction method based on road information |
| CN115188205B (en) * | 2022-07-04 | 2024-03-29 | 武汉理工大学 | Road information-based automobile driving condition correction method |
| CN115447554A (en) * | 2022-10-28 | 2022-12-09 | 北京易控智驾科技有限公司 | Method and device for detecting braking performance of unmanned mining vehicle, electronic equipment and storage medium |
| WO2024087447A1 (en) * | 2022-10-28 | 2024-05-02 | 北京易控智驾科技有限公司 | Method and apparatus for testing brake performance of unmanned mine vehicle, electronic device, storage medium, and computer program product |
| CN116776229A (en) * | 2023-08-17 | 2023-09-19 | 深圳市城市交通规划设计研究中心股份有限公司 | Method for dividing typical running conditions of automobile for carbon factor emission |
| CN116776229B (en) * | 2023-08-17 | 2023-12-26 | 深圳市城市交通规划设计研究中心股份有限公司 | Method for dividing typical running conditions of automobile facing carbon emission factors |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113689594A (en) | Novel automobile driving condition construction method | |
| CN111832225B (en) | Method for constructing driving condition of automobile | |
| CN107067722B (en) | Method for constructing vehicle running condition | |
| WO2020244288A1 (en) | Method and apparatus for evaluating truck driving behaviour based on gps trajectory data | |
| CN108198425A (en) | A kind of construction method of Electric Vehicles Driving Cycle | |
| CN109840612A (en) | User's driving behavior analysis method and system | |
| CN109543874B (en) | Airport air quality prediction method combining meteorological condition influence | |
| CN110826785B (en) | High-risk road section identification method based on k-medoids clustering and Poisson inverse Gaussian | |
| CN113297795A (en) | Method for constructing running condition of pure electric vehicle | |
| JP7331184B2 (en) | Construction Method of Standard Driving Mode for Heavy-Duty Trucks Including Road Gradient Information | |
| CN110836675A (en) | Decision tree-based automatic driving search decision method | |
| CN113222385A (en) | Method for constructing and evaluating driving condition of electric automobile | |
| CN109408955B (en) | Energy consumption analysis method and system for electric automobile | |
| CN116739376A (en) | Highway pavement preventive maintenance decision method based on data mining | |
| CN112948965A (en) | Method for constructing automobile driving condition based on machine learning and statistical verification | |
| CN114970943A (en) | Expressway accident type and severity prediction method | |
| CN110459067B (en) | Traffic green road signal coordination control evaluation method and system based on vehicle individuals | |
| CN115718886A (en) | Construction method and device for pure electric vehicle running condition | |
| Guo et al. | Application of PCA-K-means++ combination model to construction of light vehicle driving conditions in intelligent traffic | |
| CN113744530B (en) | Construction method of vehicle working condition | |
| JP2023164240A (en) | Method for designing vehicle speed/slope compound operation condition cycle test | |
| CN114841283A (en) | Method, device, equipment and medium for determining running condition of new energy vehicle | |
| CN114013285B (en) | Actual driving range evaluation method for electric automobile | |
| CN112948752B (en) | Typical working condition construction method for long movement segment of train | |
| CN115271001B (en) | Vehicle driving condition identification method and device, vehicle and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211123 |