CN113297795A - Method for constructing running condition of pure electric vehicle - Google Patents
Method for constructing running condition of pure electric vehicle Download PDFInfo
- Publication number
- CN113297795A CN113297795A CN202110595392.0A CN202110595392A CN113297795A CN 113297795 A CN113297795 A CN 113297795A CN 202110595392 A CN202110595392 A CN 202110595392A CN 113297795 A CN113297795 A CN 113297795A
- Authority
- CN
- China
- Prior art keywords
- electric vehicle
- pure electric
- data
- running
- short
- 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
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000012360 testing method Methods 0.000 claims abstract description 51
- 238000010276 construction Methods 0.000 claims abstract description 16
- 101000822633 Pseudomonas sp 3-succinoylsemialdehyde-pyridine dehydrogenase Proteins 0.000 claims abstract description 13
- 238000000513 principal component analysis Methods 0.000 claims abstract description 12
- 238000012216 screening Methods 0.000 claims abstract description 12
- 230000001133 acceleration Effects 0.000 claims description 19
- 239000012634 fragment Substances 0.000 claims description 13
- 238000007637 random forest analysis Methods 0.000 claims description 12
- 238000011156 evaluation Methods 0.000 claims description 11
- 238000003064 k means clustering Methods 0.000 claims description 11
- 238000002474 experimental method Methods 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000006911 nucleation Effects 0.000 claims description 4
- 238000010899 nucleation Methods 0.000 claims description 4
- 238000012847 principal component analysis method Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 description 9
- 239000011159 matrix material Substances 0.000 description 7
- 238000007621 cluster analysis Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 3
- 239000000446 fuel Substances 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- XNKARWLGLZGMGX-UHFFFAOYSA-N ethyl 4-(4-chloro-2-methylphenoxy)butanoate Chemical compound CCOC(=O)CCCOC1=CC=C(Cl)C=C1C XNKARWLGLZGMGX-UHFFFAOYSA-N 0.000 description 1
- 239000000295 fuel oil Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G06Q50/40—
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Educational Administration (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention discloses a method for constructing the running condition of a pure electric vehicle, which comprises the following processes of carrying out data acquisition on the running condition of the pure electric vehicle, dividing a test route into a plurality of short stroke segments, and acquiring characteristic parameters of the running condition of the pure electric vehicle from the plurality of short stroke segments; carrying out nonlinear dimensionality reduction on the characteristic parameters of the running working conditions of the pure electric vehicle through kernel principal component analysis, classifying the characteristic parameters subjected to nonlinear dimensionality reduction through a mixed clustering method, and screening a plurality of short stroke segments according to classification results and combination of the duration weight of each class in the working conditions and the Pearson correlation coefficient to construct a plurality of candidate working conditions of the pure electric vehicle; and calculating and comparing relative error values of the characteristic parameters in the plurality of candidate working conditions of the pure electric vehicle and the SAPD frequency value to construct the running working conditions of the pure electric vehicle. The construction accuracy of the working condition is higher, the actual running characteristics of the electric automobile can be reflected better, and the consistency of the obtained working condition curve and the actual working condition is stronger.
Description
Technical Field
The invention belongs to the technical field of traffic, and particularly belongs to a method for constructing a running condition of a pure electric vehicle.
Background
The driving condition of a vehicle is a speed-time curve representing a particular traffic environment, a particular vehicle driving characteristic. The running condition is a common core technology in the automobile industry and can be used for design research and development, evaluation test and the like of automobiles.
International standard working conditions are generally adopted in various vehicle test working conditions in China, and the current Chinese working conditions are synchronously promoted as reference standards but are not popularized yet. Because the running characteristics of the electric automobile and the traditional fuel automobile have great difference, and the test working condition (CLTC-P) of the Chinese light automobile represents the average characteristic of the light automobile, a targeted test working condition needs to be constructed for the pure electric automobile so as to more accurately evaluate the performances of the pure electric automobile in the aspects of energy consumption, endurance mileage, life cycle and the like.
The current working condition construction method mainly relates to a short stroke method and a Markov method. The Markov method has high complexity and non-repeatability, while the short-stroke method has simple structure, high calculation speed and wider application, but the construction precision of the Markov method depends on the precision of cluster analysis and whether a short-stroke segment for working condition synthesis is representative or not. In addition, the obtained working conditions cannot reflect the actual running characteristics of the electric automobile more truly and comprehensively due to unreasonable short-stroke segment dividing principles in the conventional short-stroke construction working condition method. In the traditional short-stroke method, the evaluation standard for the construction of the working condition is single, only one working condition curve can be synthesized through principal component analysis and cluster analysis, and the finally constructed working condition curve cannot be completely representative.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a method for constructing the running condition of a pure electric vehicle.
In order to achieve the purpose, the invention provides the following technical scheme:
a method for constructing the running condition of a pure electric vehicle comprises the following processes of carrying out data acquisition on the running condition of the pure electric vehicle, dividing a test route into a plurality of short-stroke segments, and acquiring characteristic parameters of the running condition of the pure electric vehicle from the plurality of short-stroke segments;
carrying out nonlinear dimensionality reduction on the characteristic parameters of the running working conditions of the pure electric vehicle through kernel principal component analysis, classifying the characteristic parameters subjected to nonlinear dimensionality reduction through a mixed clustering method, and screening a plurality of short stroke segments according to classification results and combination of the duration weight of each class in the working conditions and the Pearson correlation coefficient to construct a plurality of candidate working conditions of the pure electric vehicle;
and calculating and comparing relative error values of the characteristic parameters in the plurality of candidate working conditions of the pure electric vehicle and the SAPD frequency value to construct the running working conditions of the pure electric vehicle.
Preferably, the method specifically comprises the following steps,
step 3, dividing the speed and time data of the pure electric vehicle running condition into a plurality of short-stroke segments according to the running speed, and acquiring the characteristic parameters of each short-stroke segment;
step 4, carrying out nonlinear dimensionality reduction on the characteristic parameters obtained in the step 3 by a nucleation principal component analysis method, carrying out pre-classification treatment on the characteristic parameters of the short stroke segments subjected to nonlinear dimensionality reduction by a K-Means clustering method, carrying out mixed clustering analysis by combining a random forest method, screening a plurality of short stroke segments by combining the duration weight of each class in the working condition and the Pearson correlation coefficient according to classification results, and constructing a plurality of candidate working conditions of the pure electric vehicle;
and 5, determining a plurality of characteristic parameters for working condition evaluation, calculating relative error values between the plurality of characteristic parameters in the plurality of candidate working conditions of the pure electric vehicle and the running working condition data of the pure electric vehicle in the step 2, determining a typical representative working condition by combining speed-acceleration probability distribution, and constructing the running working condition of the pure electric vehicle.
Further, in the step 1, the pure electric vehicle driving data is collected through a method of combining a vehicle tracking method and vehicle-mounted measurement.
Further, in step 2, the data which are missing and abnormal in the collected automobile driving data are screened and denoised by an interpolation method and a wavelet decomposition and reconstruction method.
Preferably, in step 3, the characteristic parameters of the driving condition include a time characteristic parameter, a speed characteristic parameter and an acceleration characteristic parameter;
the time characteristic parameters comprise driving time, accelerating time, decelerating time, cruising time and parking time;
the speed characteristic parameters comprise maximum speed, average speed and speed standard deviation;
the acceleration characteristic parameters comprise maximum acceleration, average acceleration, minimum acceleration, average deceleration and standard deviation of addition and deceleration.
Preferably, in step 4, the classification categories are a low-speed category, a medium-speed category and a high-speed category.
Preferably, the formula of the Pearson correlation coefficient is as follows,
wherein X and Y represent short-stroke fragment data and total test data, r(X, Y) is the correlation coefficient of the short stroke segment and the overall experimental data,
the covariance of each point data of the short-stroke segment and each point data of the total experiment,
the average standard deviation of the data of each point of the short-stroke segment is,
the mean standard deviation of the data for each point in the overall experiment.
Preferably, the time proportion of each type of segment is calculated as follows:
in the formula: t is tiThe duration of the i-th type short stroke segment in the representative working condition to be constructed is taken as the duration;
tdrivingcycletaking the duration of the finally constructed representative working condition between 1200 and 1800 seconds;
toverallthe sum of the durations of all short stroke segments;
ti,jthe length of time occupied by the jth fragment data in the ith type short-stroke fragment is obtained;
njthe total number of segments contained in the ith short-stroke segment.
Preferably, in step 5, the characteristic parameters for evaluating the operating conditions include an average speed, an average acceleration and deceleration, time ratios of each operating state, a power demand parameter KPE of the electric vehicle, and a kinetic energy demand parameter RPA.
Preferably, in step 5, the minimum relative error value MARE between a plurality of characteristic parameters in a plurality of candidate working conditions of the pure electric vehicle and the data of the running working conditions of the pure electric vehicle in step 2 is calculated, and the SAPD frequency value in different candidate working conditions and the data of the running working conditions of the pure electric vehicle in step 2 are calculatedPercentage values of SAPD between SAPD frequency values in the datadiffCalculating the minimum relative error value MARE and the percentage value SAPDdiffAnd comparing the average values to screen out typical representative working conditions and constructing the running working conditions of the pure electric automobile.
Compared with the prior art, the invention has the following beneficial technical effects:
the invention provides a method for constructing a running condition of an electric vehicle, which realizes nonlinear dimension reduction of characteristic parameters of short-stroke segments by utilizing a kernel principal component analysis (K-PCA) method according to collected running data of the electric vehicle conforming to a special region and area, provides an improved KR mixed clustering method for constructing a candidate cycle condition, and combines a traditional K-Means clustering algorithm with a random forest algorithm for carrying out cluster analysis on the short-stroke segments. And determining an evaluation standard by considering the specific characteristic parameters and the speed-acceleration distribution probability aiming at the electric automobile, and selecting a representative cycle working condition from the candidate working conditions. Therefore, the construction accuracy of the working condition is higher, the actual running characteristics of the electric automobile can be reflected better, and the consistency of the obtained working condition curve and the actual working condition is stronger.
The method for constructing the running condition of the electric automobile can more accurately reflect the actual running characteristics of the electric automobile. Considering that the electric automobile does not have an idle speed condition in the actual running process during the short-stroke segment division, a new short-stroke segment division standard for the electric automobile is established, which is different from the traditional fuel oil automobile in that the acceleration, deceleration, constant speed and idle speed segments are taken as division bases.
The working condition construction method provided by the invention maintains the advantages of the combination of principal component analysis and K-Means clustering method in the short stroke, adopts the kernel principal component analysis to perform nonlinear dimension reduction on the characteristic parameters in consideration of the nonlinear characteristics of the driving data, and provides a more optimized mixed clustering method. The K-Means clustering result is further classified and predicted by a random forest algorithm with good classification capability and generalization capability in machine learning, the clustering result is optimized, the classification accuracy is improved, and the working condition construction precision is further improved. And compared with other classification algorithms, the random forest algorithm can effectively process data with large data sets and high feature dimensions.
The invention establishes a working condition effectiveness evaluation method aiming at the driving characteristics and the performance characteristics of an electric automobile, which mainly comprises the following steps: evaluating the clustering effect by adopting a clustering effectiveness comprehensive evaluation index to ensure the clustering effectiveness; providing a short-stroke representative fragment screening method for working condition synthesis, and constructing 10 candidate working conditions; and a candidate working condition screening and evaluating method is provided, so that a typical representative driving working condition curve which accords with the driving characteristics of the electric automobile can be finally constructed.
Drawings
FIG. 1 is a flow chart of a method for constructing a running condition of a pure electric vehicle according to the invention;
FIG. 2 is a schematic diagram of an actual test road in an embodiment of the present invention;
FIG. 3 is a comparison of vehicle speed curves before and after data preprocessing in an embodiment of the present invention;
FIG. 4 is a flow chart of a hybrid clustering algorithm in an embodiment of the present invention;
FIG. 5 is a result diagram of a hybrid clustering algorithm employed in the embodiments of the present invention;
FIG. 6 is a typical representative operating condition curve of a pure electric vehicle constructed in an embodiment of the present invention.
Detailed Description
The present invention will now be described in further detail with reference to specific examples, which are intended to be illustrative, but not limiting, of the invention.
The invention provides a construction method of running conditions of an electric automobile, which mainly comprises the following steps:
Step 1.2, determining different road lengths and proportions in the test route by adopting an analytic hierarchy process: the multi-objective decision problem is completed based on an analytic hierarchy process, and an optimal travel mode model of the driver is established. Three levels are included: (1) target layer: an optimal trip mode; (2) a criterion layer: quick travel and convenient travel serve as alternative criteria; (3) scheme layer: different road types are taken as candidate schemes, including express roads, main roads, secondary roads, branch roads and the like.
The analytic steps of the analytic hierarchy process are mainly divided into 3 steps: analyzing each element in the system and evaluating the importance; judging the relative weight of the selected elements and a certain criterion; and thirdly, sequencing the levels according to the total weight. And finally, the lengths and proportions of the roads with different grades in the test route are obtained through analysis and calculation.
And step 1.3, finishing the planning of the test route based on the total length of the test road determined in the step 1.1 and the lengths and proportions of different roads in the test route determined in the step 1.2. And carrying out a real-time test based on the planned route, and acquiring the running data of the pure electric vehicle by adopting a method combining a vehicle tracking method and vehicle-mounted measurement to obtain a series of speed and time data representing the running condition of the electric vehicle.
And 2, preprocessing the pure electric vehicle driving speed and time data collected in the step 1.
In the process of acquiring vehicle speed and time data in a real vehicle test, conditions of GPS signal loss of acquisition equipment and abnormal signal acquisition of the equipment (such as signal abnormal peak value) inevitably exist, and in order to avoid the influence of abnormal data on the construction precision of the working condition, the missing and abnormal data in the pure electric vehicle driving data acquired in the step 1 are screened and subjected to noise reduction by using an interpolation method and a wavelet decomposition and reconstruction method, so that the preprocessed vehicle driving speed and time data are obtained. I.e. the raw data hereinafter.
And 3, dividing short-stroke fragments. Dividing the series of random continuous vehicle speed and time test data obtained in the step 1 into a plurality of short stroke segments, wherein each short stroke segment is divided according to the speed, namely the short stroke segment is defined as a segment between two continuous parking segments, each short stroke segment comprises a plurality of random vehicle speed data of acceleration, deceleration, cruising and parking, and the time-related characteristic parameter, the speed-related characteristic parameter and the acceleration-related characteristic parameter of each short stroke segment are obtained. The method comprises the following specific steps:
and 3.1, based on a short-stroke method, considering the no-idle state of the pure electric vehicle, providing a speed state division principle, defining a short-stroke section as a section between two continuous parking sections based on the principle, wherein each short-stroke section consists of a plurality of random acceleration, deceleration, cruise and parking speed data, and finally dividing a series of complete continuous speed data in the original data into a plurality of short-stroke sections.
Wherein v (km/h) and a (m/s)2) Representing the speed and acceleration of the vehicle, respectively.
Step 3.2, extracting 14 characteristic parameters representing the automobile running working condition from each short stroke section according to the series of short stroke sections obtained by dividing in the step 3.1 and combining the running characteristics and the performance characteristics of the pure electric automobile, wherein the characteristic parameters are shown in table 1 and are used as characteristic bases for subsequently classifying the short stroke sections;
table 1 describes characteristic parameters of short stroke segments
Step 4, constructing candidate working conditions of the pure electric vehicle;
and 4.1, carrying out nonlinear dimensionality reduction on the short stroke segment characteristic parameters obtained in the step 3 by a kernel principal component analysis (K-PCA) method.
And 4.2, constructing candidate cycle conditions for the short stroke segment characteristic parameters subjected to nonlinear dimensionality reduction by a hybrid clustering method.
Specifically, the short stroke segment characteristic parameters subjected to nonlinear dimensionality reduction are pre-classified by adopting K-Means clustering, a training set is selected from various K-Means clustering results by combining a random forest method, and the residual driving data of various categories are classified and predicted. Selecting Compactness (CP), separation degree (SP), Davies-Bouldin index (DB) and Dunn Validity Index (DVI) as clustering effectiveness evaluation indexes, namely evaluating whether each type of short-stroke fragment data can be accurately classified into a specific type or not so as to ensure the clustering effectiveness, ensuring that the characteristic difference of the data among the types is large and the characteristic difference of the data in each type is small so as to quantitatively evaluate the clustering effect.
And 4.3, after cluster analysis, dividing the short stroke segment library into a plurality of specific categories according to different categories of vehicle speed characteristics, splicing and combining the short stroke segments based on the Pearson correlation coefficient and the time length proportion occupied by each category of short stroke segments, and finally constructing 10 candidate working conditions.
The larger the correlation coefficient of the short stroke segment with the overall test data is, the more the short stroke segment can represent the overall characteristics of the belonging category. The formula for the pearson correlation coefficient is as follows: based on the formula, 10 segments with the maximum correlation coefficient with the total test data are respectively screened from various short-stroke segment libraries obtained after clustering, and are used as various representative short-stroke segments for recombination of subsequent short-stroke segments to generate a representative working condition curve.
Wherein X and Y represent short-stroke segment data and total test data, respectively, r (X, Y) is a correlation coefficient of the short-stroke segment and the total test data,
the covariance of each point data of the short-stroke segment and each point data of the total experiment,
the average standard deviation of the data of each point of the short-stroke segment is,
the mean standard deviation of the data for each point in the overall experiment.
Setting a total working condition duration, and splicing and combining the short stroke segments according to the duration proportion of each type of short stroke segments, wherein the time proportion of each type of segments is calculated as follows:
in the formula: t is tiThe duration of the i-th type short stroke segment in the representative working condition to be constructed is taken as the duration;
tdrivingcyclethe duration of the finally constructed representative working condition is generally between 1200 and 1800 seconds;
toverallthe sum of the durations of all short stroke segments;
ti,jthe length of time occupied by the jth fragment data in the ith type short-stroke fragment is obtained;
njthe total number of segments contained in the ith short-stroke segment.
Step 5, constructing a representative running working condition of the pure electric vehicle;
and 5.1, selecting 10 characteristic parameters for working condition evaluation through statistical analysis, wherein the 10 characteristic parameters comprise average speed, average acceleration and deceleration, time proportion of each running state, power demand parameter KPE and kinetic energy demand parameter RPA of the electric automobile.
And 5.2, calculating relative error values (ARE) of 10 characteristic parameters between the 10 candidate working condition data and the vehicle speed data preprocessed in the step 2, and calculating a minimum relative error value (MARE) by taking the candidate working condition of which the relative error value of each characteristic parameter value is less than 10%. Calculating the percentage value SAPD between the SAPD frequency value in different candidate working conditions and the SAPD frequency value in the original data by combining the velocity-acceleration probability distribution (SAPD)diff。
Step 5.3, calculate MARE and SAPDdiffThe average value of the PV values is used as a candidate working condition screening index PV value, and the candidate working condition with the minimum PV value is taken as the most typical representative working condition.
Examples
In detail, the present invention will be further described with reference to the accompanying drawings and specific embodiments. Example (b): taking a certain large-scale first-line city in China as an example, the running condition of the pure electric automobile is constructed.
Step 1: and acquiring running condition data.
By combining the factors of road proportion of each level, road traffic flow of each level, automobile holding capacity, citizen's trip road selection will and the like in the area, an analytic hierarchy process is adopted to complete test planning, and a test route is designed as shown in figure 2. The length of the designed test road is 38.46km, and the proportions of the express way, the main road, the secondary road and the branch in the test route are 29.96%, 24.80%, 26.85% and 18.39% respectively. Selecting a pure electric vehicle with a large reserved quantity in a city as a test vehicle by adopting a method of combining a vehicle tracking method and vehicle-mounted measurement; selecting 28 experienced drivers to circularly drive and collect data on a test route in the early peak period of 7:00-9:00, the late peak period of 12:30-14:30 and the late peak period of 19:00-21:00 of a city; the main test equipment comprises a GPS, an OBD and a VBOX, and is mainly used for collecting a series of random continuous speed and time data of automobile running.
Step 2: and (4) data preprocessing, namely screening, denoising and smoothing the original data.
And (4) performing data interpolation by using the following function to remove singular points in the data. In the formula vtVehicle speed, v 'representing the current point in time'tRepresenting the vehicle speed at the current point in time after processing.
In order to prevent the short-stroke segment division from being too disordered and prevent the segment length from being too small, the noise reduction smoothing processing is carried out by utilizing a wavelet decomposition and reconstruction method. FIG. 3 shows a comparison of partial velocity profiles before and after pretreatment.
And step 3: and dividing short-stroke fragments.
And (3) dividing a series of complete and continuous vehicle speed data obtained in the step (2) into a plurality of short-stroke segments. Different from the traditional fuel vehicle, the traditional fuel vehicle usually takes an idling state as a division point between two short stroke segments, and considering the no-idling state of the pure electric vehicle, by adopting the method provided by the invention, one short stroke is defined as a segment between two continuous parking segments, namely, two adjacent points with the vehicle speed of 0 are taken as the basis for dividing the vehicle speed segment, a plurality of random parking, accelerating, decelerating and cruising processes are included between the two continuous parking segments, and meanwhile, the short stroke segment with the running time of less than 10 seconds is excluded, so that 1414 effective short stroke segments are obtained.
Wherein v (km/h) and a (m/s)2) Representing the speed and acceleration of the vehicle, respectively.
According to 1414 short-stroke segments obtained by division, short-stroke segment characteristic parameter values are calculated to obtain a short-stroke segment data characteristic matrix 1414 × 14, and the result is shown in table 2.
TABLE 2 short Stroke fragment characteristic parameter values
And 4, step 4: constructing a candidate working condition curve of the pure electric vehicle;
in detail, the steps of constructing the candidate working condition of the pure electric automobile are as follows:
step 4.1, nucleation principal component analysis (K-PCA): and extracting characteristic parameters, performing nonlinear dimension reduction on the data to reduce characteristic redundancy, and calculating the characteristic parameter values of the short-stroke segment of the method in the table 2. However, the principal component analysis adopted in the conventional method can only extract the linear features of the data, and the nonlinear problem widely exists and cannot be fully solved. Therefore, the invention proposes to adopt a nucleation principal component analysis method.
The invention provides a specific step for reducing short-stroke characteristic parameter dimension by using KPCA: (1) to eliminate the effect of the magnitude on the result, the table 21414 × 14 characteristic parameter matrix is normalized; (2) calculating a kernel matrix: using a gaussian radial basis kernel function:
wherein, σ represents the width of the Gaussian function, and a relatively suitable parameter σ is obtained through parameter optimization, wherein the parameter σ is 38.73; k is 1,2 … M, K is an M × M kernel matrix, M is 1414;
(3) solving a characteristic value and a characteristic vector of K by adopting a Singular Value Decomposition (SVD) algorithm;
(4) arranging the eigenvalues in descending order (lambda)1≥λ2≥…≥λM) For the first p nonzero numbersThe feature vector corresponding to the feature value is normalized according to the following formula:
i.e. in the feature space F, λ must be madek(ak·αk) 1. Wherein λ iskRepresenting the kth non-zero eigenvalue, the corresponding K eigenvector being arranged as alpha according to its eigenvalue1,…,αp,…αM。
(5) The projection value (principal component) of one test sample x on the feature vector is calculated and used as a new feature:
for a new sample x, extracting its principal component only needs to map the corresponding mapped sample phi (x) in its F space to VkAnd (5) projection is carried out.
(6) And calculating to obtain each principal component and the accumulated contribution rate thereof according to the steps, taking the first principal components with the accumulated contribution rate of 85% as the basis of subsequent clustering analysis, and finally calculating a principal component score matrix as the variable of an output space.
TABLE 3 principal Components and cumulative contribution rates
The principal component analysis results are shown in table 3, and after K-PCA analysis, 4 principal components were obtained, with a cumulative variance contribution of 91.96%. For comparison, 3 principal components were obtained by PCA principal component analysis, and the cumulative variance contribution rate was 85.49%. Therefore, the K-PCA adopted by the invention carries out nonlinear dimensionality reduction on the short-stroke fragment data, and the obtained low-dimensional data contains more effective original data information.
And 4.2, performing cluster analysis.
The KR mixed clustering algorithm implementation flow chart of the K-Means combined random forest provided by the invention is shown in the attached figure 4, and comprises the following specific steps:
(1) and substituting the original data by the coring principal component score matrix, and classifying the coring principal component score matrix into three categories of low speed, medium speed and high speed through K-Means clustering.
(2) And taking the 30 short stroke segments closest to the clustering center in the three K-Means results as random forest training samples, and completing the classification prediction of the remaining short stroke segments.
TABLE 4 Classification of feature results
The classification results of K-Means clustering combined with random forests (K-Means + random forests) and K-Means clustering are shown in FIG. 5 and Table 4. After being clustered by partial K-Means, the short-stroke segments belonging to the medium-speed class are classified into the first class after random forest classification, so that the clustering result is optimized, and the clustering accuracy is improved.
(3) Table 5 shows the classification effectiveness evaluation indexes of the K-Means and mixed clustering (K-Means + random forest) method. Compared with K-Means clustering, CP and DB of mixed clustering are smaller, which indicates that the intra-class distance is closer; whereas the larger SP and DVI values indicate the farther the inter-class distance. The result shows that the hybrid clustering algorithm provided by the invention can effectively improve the similarity in the same cluster and reduce the similarity among different clusters.
TABLE 5 evaluation results of Classification effectiveness
And 4.3, constructing candidate working conditions.
As shown in FIG. 6, a representative cycle duration is set to 1200s, and the representative cycle duration is composed of three classes of short stroke segments which are combined according to time proportion, and the calculation result shows that the low speed class is 160s, the medium speed class is 603s, and the high speed class is 437 s. And selecting the represented short stroke segments according to the Pearson correlation coefficients of the short stroke segments and the total test data, splicing and combining to finally construct 10 candidate working conditions.
And 5: and constructing a typical representative working condition curve of the pure electric vehicle.
Step 5.1, the data characteristics of the cycle conditions characterized by 10 characteristic parameters are shown in table 6.
And 5.2, screening a typical representative working condition from 10 candidate working conditions by using the principle of minimum PV value according to the candidate working condition screening index PV value calculation method provided by the invention.
The results of comparison of representative operating condition characteristics constructed based on the conventional short stroke method and the method proposed by the present invention are shown in table 6. The PV values between the constructed representative condition and the raw driving data for development and verification are 2.72% and 2.95%, respectively, while the PV values between the constructed condition and the raw data of the conventional short stroke method are 3.64% and 3.81%, respectively. The working condition construction method provided by the invention can effectively reflect the driving characteristics of the pure electric vehicle to be researched, has higher working condition construction precision and better consistency with the actual driving working condition, and proves the effectiveness and reliability of the method.
Table 6 represents the comparison of the characteristic parameters of the working conditions
Claims (10)
1. A method for constructing the running condition of a pure electric vehicle is characterized by comprising the following steps of carrying out data acquisition on the running condition of the pure electric vehicle, dividing a test route into a plurality of short-stroke segments, and acquiring characteristic parameters of the running condition of the pure electric vehicle from the plurality of short-stroke segments;
carrying out nonlinear dimensionality reduction on the characteristic parameters of the running working conditions of the pure electric vehicle through kernel principal component analysis, classifying the characteristic parameters subjected to nonlinear dimensionality reduction through a mixed clustering method, and screening a plurality of short stroke segments according to classification results and combination of the duration weight of each class in the working conditions and the Pearson correlation coefficient to construct a plurality of candidate working conditions of the pure electric vehicle;
and calculating and comparing relative error values of the characteristic parameters in the plurality of candidate working conditions of the pure electric vehicle and the SAPD frequency value to construct the running working conditions of the pure electric vehicle.
2. The method for constructing the running condition of the pure electric vehicle according to claim 1, characterized by comprising the following steps,
step 1, determining the total length of a test route, determining the lengths and proportions of different roads in the test route by adopting an analytic hierarchy process, planning the test route according to the total length of the test route and the lengths and proportions of the different roads in the test route, and acquiring data of the running condition of the pure electric vehicle on the planned test route to obtain speed and time data of the running condition of the pure electric vehicle;
step 2, carrying out data preprocessing on the speed and time data of the pure electric vehicle driving condition collected in the step 1;
step 3, dividing the speed and time data of the pure electric vehicle running condition into a plurality of short-stroke segments according to the running speed, and acquiring the characteristic parameters of each short-stroke segment;
step 4, carrying out nonlinear dimensionality reduction on the characteristic parameters obtained in the step 3 by a nucleation principal component analysis method, carrying out pre-classification treatment on the characteristic parameters of the short stroke segments subjected to nonlinear dimensionality reduction by a K-Means clustering method, carrying out mixed clustering analysis by combining a random forest method, screening a plurality of short stroke segments by combining the duration weight of each class in the working condition and the Pearson correlation coefficient according to classification results, and constructing a plurality of candidate working conditions of the pure electric vehicle;
and 5, determining a plurality of characteristic parameters for working condition evaluation, calculating relative error values between the plurality of characteristic parameters in the plurality of candidate working conditions of the pure electric vehicle and the running working condition data of the pure electric vehicle in the step 2, determining a typical representative working condition by combining speed-acceleration probability distribution, and constructing the running working condition of the pure electric vehicle.
3. The pure electric vehicle driving condition construction method according to claim 2, characterized in that in step 1, pure electric vehicle driving data are collected by a method combining a vehicle tracking method and vehicle-mounted measurement.
4. The pure electric vehicle driving condition construction method according to claim 2, characterized in that in step 2, data which are missing and abnormal in the acquired vehicle driving data are subjected to screening and denoising treatment by using an interpolation method and a wavelet decomposition and reconstruction method.
5. The method for constructing the running condition of the pure electric vehicle according to claim 1, wherein in step 3, the characteristic parameters of the running condition comprise a time characteristic parameter, a speed characteristic parameter and an acceleration characteristic parameter;
the time characteristic parameters comprise driving time, accelerating time, decelerating time, cruising time and parking time;
the speed characteristic parameters comprise maximum speed, average speed and speed standard deviation;
the acceleration characteristic parameters comprise maximum acceleration, average acceleration, minimum acceleration, average deceleration and standard deviation of addition and deceleration.
6. The method for constructing the running condition of the pure electric vehicle according to claim 1, wherein in step 4, the classification categories are a low-speed category, a medium-speed category and a high-speed category.
7. The pure electric vehicle driving condition construction method according to claim 1, wherein the formula of the Pearson correlation coefficient is as follows,
wherein X and Y represent short-stroke segment data and total test data, respectively, r (X, Y) is a correlation coefficient of the short-stroke segment and the total test data,
the covariance of each point data of the short-stroke segment and each point data of the total experiment,
the average standard deviation of the data of each point of the short-stroke segment is,
the mean standard deviation of the data for each point in the overall experiment.
8. The method for constructing the running condition of the pure electric vehicle according to claim 1, wherein the time proportion of each type of segment is calculated according to the following formula:
in the formula: t is tiThe duration of the i-th type short stroke segment in the representative working condition to be constructed is taken as the duration;
tdrivingcycletaking the duration of the finally constructed representative working condition between 1200 and 1800 seconds;
toverallthe sum of the durations of all short stroke segments;
ti,jthe length of time occupied by the jth fragment data in the ith type short-stroke fragment is obtained;
njthe total number of segments contained in the ith short-stroke segment.
9. The pure electric vehicle running condition construction method according to claim 1, wherein in step 5, the characteristic parameters for evaluating the working conditions comprise an average speed, an average acceleration and deceleration, time proportions of each running state, a power demand parameter KPE and a kinetic energy demand parameter RPA of the electric vehicle.
10. The method for constructing the running condition of the pure electric vehicle as claimed in claim 1, wherein in step 5, a minimum relative error value MARE between a plurality of characteristic parameters in a plurality of candidate conditions of the pure electric vehicle and the running condition data of the pure electric vehicle in step 2 is calculated, and a percentage value SAPD between a SAPD frequency value in different candidate conditions and a SAPD frequency value in the running condition data of the pure electric vehicle in step 2 is calculateddiffCalculating the minimum relative error value MARE and the percentage value SAPDdiffAnd comparing the average values to screen out typical representative working conditions and constructing the running working conditions of the pure electric automobile.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110595392.0A CN113297795A (en) | 2021-05-28 | 2021-05-28 | Method for constructing running condition of pure electric vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110595392.0A CN113297795A (en) | 2021-05-28 | 2021-05-28 | Method for constructing running condition of pure electric vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113297795A true CN113297795A (en) | 2021-08-24 |
Family
ID=77326069
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110595392.0A Pending CN113297795A (en) | 2021-05-28 | 2021-05-28 | Method for constructing running condition of pure electric vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113297795A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113963462A (en) * | 2021-12-16 | 2022-01-21 | 中汽研汽车检验中心（天津）有限公司 | Electric drive assembly reliability working condition construction method and construction device based on remote transmission data |
| CN113984406A (en) * | 2021-10-26 | 2022-01-28 | 长安大学 | Short-time working condition construction method and system for electric vehicle safety rapid detection |
| CN114084150A (en) * | 2021-12-22 | 2022-02-25 | 中国北方车辆研究所 | Method for generating running condition of off-road vehicle |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198425A (en) * | 2018-02-10 | 2018-06-22 | 长安大学 | A kind of construction method of Electric Vehicles Driving Cycle |
| CN110717147A (en) * | 2019-10-10 | 2020-01-21 | 辽宁工程技术大学 | Method for constructing driving condition of automobile |
-
2021
- 2021-05-28 CN CN202110595392.0A patent/CN113297795A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198425A (en) * | 2018-02-10 | 2018-06-22 | 长安大学 | A kind of construction method of Electric Vehicles Driving Cycle |
| CN110717147A (en) * | 2019-10-10 | 2020-01-21 | 辽宁工程技术大学 | Method for constructing driving condition of automobile |
Non-Patent Citations (2)
| Title |
|---|
| LU WANG等: "Development of a Typical Urban Driving Cycle for Battery Electric Vehicles Based on Kernel Principal Component Analysis and Random Forest", 《IEEE ACCESS》 * |
| 余曼等: "基于FCM聚类算法的电动车城市循环工况构建", 《公路交通科技》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113984406A (en) * | 2021-10-26 | 2022-01-28 | 长安大学 | Short-time working condition construction method and system for electric vehicle safety rapid detection |
| CN113984406B (en) * | 2021-10-26 | 2023-07-14 | 长安大学 | Short-time working condition construction method and system for electric automobile safety rapid detection |
| CN113963462A (en) * | 2021-12-16 | 2022-01-21 | 中汽研汽车检验中心（天津）有限公司 | Electric drive assembly reliability working condition construction method and construction device based on remote transmission data |
| CN113963462B (en) * | 2021-12-16 | 2022-02-25 | 中汽研汽车检验中心（天津）有限公司 | Electric drive assembly reliability working condition construction method and construction device based on remote transmission data |
| CN114084150A (en) * | 2021-12-22 | 2022-02-25 | 中国北方车辆研究所 | Method for generating running condition of off-road vehicle |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Development of a representative urban driving cycle construction methodology for electric vehicles: A case study in Xi’an | |
| CN113297795A (en) | Method for constructing running condition of pure electric vehicle | |
| CN110414584B (en) | Automobile road spectrum clustering synthesis method and system | |
| CN109552338B (en) | Evaluation method and system for ecological driving behavior of pure electric vehicle | |
| CN108198425A (en) | A kind of construction method of Electric Vehicles Driving Cycle | |
| CN106203856A (en) | A kind of Combined Principal Components analysis and the vehicle driving-cycle formulating method of Fuzzy c-means Clustering | |
| CN108647836B (en) | Driver energy-saving evaluation method and system | |
| CN113222385B (en) | Method for constructing and evaluating driving condition of electric automobile | |
| CN113689594A (en) | Novel automobile driving condition construction method | |
| CN110836675A (en) | Decision tree-based automatic driving search decision method | |
| CN111563555A (en) | Driver driving behavior analysis method and system | |
| CN113581188A (en) | Commercial vehicle driver driving style identification method based on Internet of vehicles data | |
| CN113688558A (en) | Automobile driving condition construction method and system based on large database samples | |
| CN113232669A (en) | Driving style identification method based on machine learning | |
| CN115935672A (en) | Fuel cell automobile energy consumption calculation method fusing working condition prediction information | |
| CN115422747A (en) | Method and device for calculating discharge amount of pollutants in tail gas of motor vehicle | |
| CN114971156A (en) | Driving style analysis method based on big data dangerous scene | |
| Wang et al. | Development of a typical urban driving cycle for battery electric vehicles based on kernel principal component analysis and random forest | |
| CN110796374A (en) | Construction method of urban comprehensive driving working condition with typical lane changing and turning working conditions | |
| CN114048798A (en) | Automobile driving condition construction method based on improved noise reduction self-encoder | |
| 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 | |
| CN112035536A (en) | Electric automobile energy consumption prediction method considering dynamic road network traffic flow | |
| CN115774942A (en) | Driving style identification model modeling and statistical method based on Internet of vehicles real vehicle data and SVM |
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 |