CN115264048B - Intelligent gear decision design method for automatic transmission based on data mining - Google Patents
Intelligent gear decision design method for automatic transmission based on data mining Download PDFInfo
- Publication number
- CN115264048B CN115264048B CN202210882192.8A CN202210882192A CN115264048B CN 115264048 B CN115264048 B CN 115264048B CN 202210882192 A CN202210882192 A CN 202210882192A CN 115264048 B CN115264048 B CN 115264048B
- Authority
- CN
- China
- Prior art keywords
- data
- gear
- driving
- working condition
- automatic transmission
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
- F16H—GEARING
- F16H61/00—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing
- F16H61/02—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing characterised by the signals used
- F16H61/0202—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing characterised by the signals used the signals being electric
- F16H61/0204—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing characterised by the signals used the signals being electric for gearshift control, e.g. control functions for performing shifting or generation of shift signal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F16—ENGINEERING ELEMENTS AND UNITS; GENERAL MEASURES FOR PRODUCING AND MAINTAINING EFFECTIVE FUNCTIONING OF MACHINES OR INSTALLATIONS; THERMAL INSULATION IN GENERAL
- F16H—GEARING
- F16H61/00—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing
- F16H2061/0075—Control functions within control units of change-speed- or reversing-gearings for conveying rotary motion ; Control of exclusively fluid gearing, friction gearing, gearings with endless flexible members or other particular types of gearing characterised by a particular control method
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Mechanical Engineering (AREA)
- Quality & Reliability (AREA)
- Control Of Transmission Device (AREA)
Abstract
The invention discloses a data mining-based intelligent gear decision design method for an automatic transmission, which comprises the following steps: acquiring driving data of a good driver driving a manual transmission vehicle in various road environments; preprocessing the collected driving data; data cleaning is carried out on the preprocessed driving data, and gear decision strategies of excellent drivers are obtained by dividing various driving working conditions and respectively constructing the gear decision strategies; three gear shifting features under a target working condition are selected, a driving data set is constructed, outliers in the data set are detected by using a support vector machine algorithm, and the outliers are removed; extracting a gear shifting boundary point of each gear data in the data set by using an Alpha-shapes algorithm; and fitting the extracted gear shifting boundary points by using a moving least square method to obtain a gear shifting regular curved surface. The gear decision is constructed from the driving data of the manual gear vehicle driven by the excellent driver and applied to the automatic car stop, so that the adaptability of the automatic car stop to the driving intention and the driving environment of the driver is improved.
Description
Technical Field
The invention relates to the technical field of automatic transmission control, in particular to an intelligent gear decision design method of an automatic transmission based on data mining.
Background
The automatic transmission controls a gear shifting executing mechanism of the automatic transmission to switch gears through a transmission control unit (Transmission Control Unit, TCU), so that the operation difficulty and the driving burden of a driver are greatly reduced. The shift control of an automatic transmission is a core technology of the automatic transmission, and can be divided into two layers: top layer control and bottom layer control. The bottom layer control controls the combination of the clutch to be separated by controlling the gear shifting executing mechanism of the automatic transmission, so that the gear shifting is realized, the top layer control is used for making gear decision (upshift, downshift or hold) according to the vehicle state parameters and considering the intention of a driver, and therefore, the gear decision directly determines the fuel economy, the power performance and the adaptability to the driving intention and the driving environment of the automatic transmission vehicle.
Existing gear decision making methods can be broadly divided into two main categories: a gear decision method based on performance index optimization (such as optimal economy, optimal dynamic performance, optimal global performance and the like) and a gear decision method based on intelligent control (such as a modified gear decision method, a learning gear decision method, a self-adaptive online optimization gear decision method and the like). The two methods mostly adopt modern control means such as dynamic programming, model prediction, a neural network, a genetic algorithm, fuzzy control and the like, take more influencing factors related to a human-vehicle-road environment into consideration, and carry out correction optimization on the basis of a traditional standard two/three-parameter gear shifting strategy to obtain gear decisions, wherein the effects are largely determined by the rationality and accuracy of an engine dynamic model and a control optimization model, and the influence components of own experience and design level of a designer are more, so that the problem that the decision gear of an automatic speed change automobile is inconsistent with the gear expected value of a user frequently occurs.
Therefore, the prior art has the defects that an intelligent gear decision design method of an automatic transmission based on data mining is lack, and the gear decision strategy of an excellent driver is constructed from driving data obtained by driving a manual gear vehicle by the excellent driver by using the data mining method and is applied to the same type of automatic gear vehicle, so that the adaptability of the automatic transmission vehicle to the driving intention and the driving environment of the driver is improved,
disclosure of Invention
In view of at least one defect of the prior art, the invention aims to provide an intelligent gear decision design method of an automatic transmission based on data mining, which is used for constructing a gear decision strategy of an excellent driver from driving data obtained by driving a manual gear vehicle by the excellent driver by using the data mining method and applying the gear decision strategy to the same type of automatic gear vehicle, thereby improving the adaptability of the automatic transmission vehicle to the driving intention and the driving environment of the driver and realizing the unification of the expected gear of the driver and the decision gear of the vehicle.
In order to achieve the above purpose, the invention adopts the following technical scheme: an intelligent gear decision design method of an automatic transmission based on data mining comprises the following steps:
step S1: acquiring driving data of a manual transmission vehicle driven by an excellent driver in various road environments, wherein the acquired driving conditions mainly comprise urban roads, suburban roads, mountain roads and urban express ways;
step S2: preprocessing the acquired driving data, including data synchronization, data resampling, data filtering and missing value processing;
step S3: the method comprises the steps of performing data cleaning on the preprocessed driving data, extracting driving data under a target working condition, constructing gear decision sub-working conditions, and obtaining a complete excellent driver gear decision strategy by dividing various driving working conditions and respectively constructing gear decision strategies;
step S4: three gear shifting features under a target working condition are selected, a driving data set is constructed, outliers in the data set are detected by using a support vector machine algorithm, and the outliers in the data set are removed;
step S5: extracting a gear shifting boundary point of each gear data in the data set by using an Alpha-shapes algorithm;
step S6: and fitting the extracted gear shifting boundary points by using a mobile least square algorithm to obtain a gear shifting regular curved surface of each gear.
In the step S2, the data resampling includes selecting a reference frequency, and resampling data of other frequencies by the reference frequency to ensure the unification of sampling frequencies; carrying out synchronization and resampling processing on the data by adopting a cubic spline curve interpolation algorithm; filtering noise of an original driving data set by adopting a Butterworth filter; defining data with zero speed and duration exceeding T seconds in the original driving data set as a missing value, wherein T is more than or equal to 2; and processing the missing value by adopting a method for deleting the original driving data in the time period of the missing value.
Because sampling starting time points are not uniform due to different delays of sensors in a data acquisition test, acquired signals such as a vehicle speed, a gear and the like are dislocated, and synchronous processing is needed for the starting time points of the dislocated signals to be aligned. In addition, since the data at the same time have no corresponding relation problem due to the different sampling frequencies of the sensors, for example, the sampling frequency of the acceleration sensor is 50Hz and the sampling frequency of the vehicle speed is 100Hz, the 10 th data in 1 second corresponds to the acceleration of 0.2 second and the vehicle speed of 0.1 second. Therefore, it is necessary to select one reference frequency to resample data of other frequencies to ensure uniformity of sampling frequencies. Since 100Hz data is the most, resampling is performed using 100Hz as the reference frequency. The invention adopts a cubic spline curve interpolation algorithm to carry out synchronous and resampling processing on data.
Because of the influence of the acquisition precision of the sensor and noise in the running environment, the original data set usually has high-frequency noise, and in order to ensure the subsequent data analysis precision, the invention adopts a Butterworth filter to filter the noise of the original data set.
The data obtained is a set of zero values due to a failure of the sampling device or a long parking time but a continuous operation of the sampling device, resulting in a certain period of data not being recorded, wherein the data of zero vehicle speed and a duration exceeding two seconds in the original data are defined as missing values. These missing values in the data set do not contain shift information of an excellent driver, but only increase the calculation amount of data processing, so that the missing values are processed by adopting a method of deleting the data in the time period in which the missing values are located.
The step S2 further includes preprocessing the driving data of the excellent driver, where the preprocessing is performed according to the following steps:
step S21: exporting the acquired driving data files into excel files, wherein the size of each excel file is not more than 120MB, determining the time starting point of each signal in each excel file, if the time starting points are inconsistent, selecting the time with the maximum time starting point as a unified time starting point, and taking the minimum ending time point of each signal as a unified ending point;
step S22: after determining a starting time point and an ending time point in each excel file, interpolating by using a cubic spline interpolation algorithm at intervals of 0.01s so as to obtain running data with the sampling frequency of 100Hz, thereby completing data synchronization and resampling in data preprocessing;
step S23: the frequency response of the butterworth filter is shown as follows:
in the formula (1), N is the order of the filter, ω represents the circular frequency, ω p For passband edge frequency omega c For a 3dB passband cut-off frequency, ε represents the normalized tuning parameter, satisfying:
wherein alpha is p Is the passband maximum attenuation coefficient; will |H (j omega) | 2 The form rewritten as a laplace function is:
wherein, the expression of the transfer function H(s) is:
wherein s is k Representing the pole, let p=s/ω c The above equation is normalized with a 3dB cut-off frequency, which is converted to a transfer function H (p):
wherein p is k Represents a normalized pole; by passband edge frequency omega p Stop band cut-off frequency omega s Maximum attenuation coefficient alpha of stop band s The filter order may be determined:
after the order of the filter is determined, a normalized pole can be obtained, and then H(s) is obtained through de-normalization, so that the transfer function of the filter can be determined.
In step S3, performing data cleaning on the preprocessed driving data to extract driving data under the target working condition includes:
and after the missing values are removed, taking the effective data between two adjacent missing value data as a data segment, and carrying out subsequent data cleaning by judging the working condition corresponding to each data segment.
The data mining-based gear decision design method of the automatic transmission is introduced by taking an ascending working condition as an example, the working condition corresponding to the data segment is the ascending working condition, and curve data, bumpy road data and low adhesion road data in driving data are required to be removed in data cleaning.
Firstly, taking effective data between two adjacent missing value data as a data segment, and judging the vehicle running condition corresponding to each data segment by utilizing steering wheel rotation angle and road resistance coefficient (reflecting road gradient) so as to carry out subsequent data cleaning;
in the curve data cleaning process, firstly calculating the steering wheel angle absolute value of each data segment, and then regarding the data segment with the steering wheel angle absolute value being more than 35 degrees and the duration exceeding 3 seconds as curve data and removing the curve data;
the change of the road resistance coefficient is due to the change of the vehicle weight, the road gradient and the rolling resistance, the data of the road resistance coefficient larger than the rolling resistance coefficient is regarded as the ascending working condition, and the road resistance coefficient f is considered in consideration of the fluctuation of the road resistance coefficient road Data with a duration greater than 5 seconds and greater than 0.02 is considered an uphill condition, data is retained during the uphill condition, and data is removed during other conditions.
The construction of the driving data set and the removal of outliers described in the step 4 are performed as follows:
step 4-1: firstly, selecting a gear shifting control parameter under a target working condition, wherein factors influencing gear shifting of a vehicle under an ascending working condition mainly comprise a vehicle speed, an accelerator pedal opening, acceleration and gradient, the ascending working condition is further divided into a uniform ascending working condition and an accelerating ascending working condition, the uniform ascending is carried out by taking the vehicle speed, the accelerator pedal opening and a road resistance coefficient as the gear shifting parameters, and the vehicle speed, the accelerator pedal opening and the acceleration are taken as the gear shifting parameters under the accelerating ascending working condition;
step 4-2: extracting a data set of each gear composed of the speed, the acceleration and the accelerator pedal opening under the acceleration working condition from the uphill working condition data obtained after pretreatment and data cleaning;
step 4-3: the data points scattered at the outermost edge of the data set obtained after preprocessing and data cleaning are caused by abnormal operation of a driver, and the outliers in the data set are detected and processed by using a support vector machine algorithm.
In step S5, a gear shift boundary point of each gear data in the dataset is extracted by using an Alpha-shapes algorithm, a vehicle speed maximum side boundary point of each gear data along the vehicle speed increasing direction is used as an upshift boundary point, and a vehicle speed minimum side boundary point is used as a downshift boundary point.
The invention provides an intelligent gear decision design method of an automatic transmission based on data mining, which constructs a gear decision strategy of an excellent driver from driving data obtained by driving a manual gear vehicle by the excellent driver by using the data mining method and is applied to the same type of automatic gear vehicle, so that the adaptability of the automatic transmission vehicle to the driving intention and the driving environment of the driver is improved, and the unification of the expected gear of the driver and the decision gear of the vehicle is realized.
Drawings
FIG. 1 is a schematic diagram of a data acquisition route for an excellent driver according to the present invention;
FIG. 2 is a flow chart of a driving data processing process;
FIG. 3 is a flow chart of data preprocessing according to the present invention;
FIG. 4 is a diagram illustrating data synchronization according to the present invention;
FIG. 5 is a schematic diagram of data resampling according to the present invention;
FIG. 6 is a flow chart of data cleansing according to the present invention;
FIG. 7 is a schematic diagram of outlier processing according to the present invention;
FIG. 8 is a schematic diagram of a shift boundary point extraction of the present invention;
FIG. 9 is a schematic view of a regular curved surface of a gear shift boundary point fitting gear shift according to the present invention;
fig. 10 is a flow chart of the method of the present invention.
Detailed Description
The invention will be described in further detail with reference to the drawings and the specific examples.
As shown in fig. 1-10, the invention discloses an intelligent gear decision design method of an automatic transmission based on data mining, which aims to solve the problem that the existing gear decision method has poor adaptability to the driving intention and the driving environment of a driver, and utilizes the data mining method to construct a gear decision strategy of the excellent driver from driving data obtained by driving a manual gear vehicle of the excellent driver and apply the gear decision strategy to the same type of automatic gear vehicle type, so that the adaptability of the automatic transmission vehicle to the driving intention and the driving environment of the driver is improved, and the unification of the expected gear of the driver and the decision gear of the vehicle is realized. The invention is not only suitable for the automatic transmission carried by the road vehicle, but also suitable for the automatic transmission carried by the engineering vehicle.
In order to achieve the above purpose, the present invention proposes the following technical scheme:
1) Acquiring mass driving data of a manual transmission vehicle driven by an excellent driver in various road environments, wherein the acquired driving conditions mainly comprise urban roads, suburban roads, mountain roads and urban express ways;
2) Preprocessing the acquired driving data, including data synchronization, data resampling, data filtering and missing value processing;
3) The method comprises the steps of performing data cleaning on the preprocessed driving data, extracting driving data under a target working condition, constructing gear decision sub-working conditions, and obtaining a complete excellent driver gear decision strategy by dividing various driving working conditions and respectively constructing gear decision strategies;
4) Three gear shifting features under a target working condition are selected, a driving data set is constructed, outliers in the data set are detected by using a support vector machine algorithm, and the outliers in the data set are removed;
5) Extracting a gear shifting boundary point of each gear data in the data set by using an Alpha-shapes algorithm;
6) And fitting the extracted boundary points by using a mobile least square algorithm to obtain a gear shifting regular curved surface.
The driving data acquisition of the excellent driver driving manual transmission vehicle in the step 1) is carried out according to the following steps:
1-1) firstly selecting a driver with more than ten years of driving age for driving a manual transmission vehicle as an excellent driver, requiring the driver to be familiar with the gear shifting operation of the experimental vehicle before carrying out a data acquisition experiment, and ensuring that the gear shifting can be carried out according to the usual gear shifting habit of the driver in the subsequent data acquisition process, thereby acquiring more real driving data of the excellent driver;
1-2) selecting a proper data acquisition driving route, wherein the driving route needs four main driving working conditions including urban roads, suburban roads, mountain roads and urban expressways.
1-3) in the data acquisition process, each driver runs three circles on the selected driving route, so that misoperation caused by unfamiliar routes is avoided.
The preprocessing of the driving data of the excellent driver in the step 2) is performed according to the following steps:
2-1) exporting the acquired driving data files into excel files, wherein the size of each excel file is not more than 120MB (the file is too large and leads to the difficulty of processing data by a general computer), determining the time starting point of each signal in each excel file, if the time starting points are inconsistent, selecting the time with the maximum time starting point as a unified time starting point, and taking the minimum ending time point of each signal as a unified ending point;
2-2) after determining a starting time point and an ending time point in each excel file, interpolating by using a cubic spline interpolation algorithm at intervals of 0.01s so as to obtain running data with the sampling frequency of 100Hz, thereby completing data synchronization and resampling in data preprocessing;
2-3) the running data after data synchronization and resampling also contains high-frequency noise components, and filters are required to be respectively designed for each signal containing noise for filtering. The frequency response of the designed butterworth filter is shown as follows:
wherein N is the order of the filter, ω represents the circular frequency, ω p For passband edge frequency omega c For a 3dB passband cut-off frequency, ε represents the normalized tuning parameter, satisfying:
wherein alpha is p Is the passband maximum attenuation coefficient. Will |H (j omega) | 2 The form rewritten as a laplace function is:
wherein, the expression of the transfer function H(s) is:
wherein s is k Representing the pole, let p=s/ω c The above equation is normalized by using a 3dB cut-off frequency, and is converted into:
wherein p is k Representing a normalized pole. By passband edge frequency omega p Stop band cut-off frequency omega s Maximum attenuation coefficient alpha of stop band s The filter order may be determined:
after the order of the filter is determined, a normalized pole can be obtained, and then H(s) is obtained through de-normalization, so that the transfer function of the filter can be determined.
The data cleaning of the driving data of the excellent driver in the step 3) is carried out according to the following steps:
3-1) firstly, taking effective data between two adjacent missing value data as a data segment, and judging the vehicle running condition corresponding to each data segment by utilizing parameters such as steering wheel rotation angle, road resistance coefficient and the like so as to carry out subsequent data cleaning;
3-2) in the curve data cleaning process, firstly calculating the steering wheel angle absolute value of each data segment, and then regarding the data segment with the steering wheel angle absolute value being more than 35 degrees and the duration exceeding 3 seconds as curve data and removing the curve data;
3-3) the change in road resistance coefficient may be due to the change in vehicle weight, road gradient and rolling resistance, uniformly considering data of road resistance coefficient larger than the rolling resistance coefficient as an uphill condition, considering the fluctuation of the road resistance coefficient, and calculating the road resistance coefficient f road Data with the duration of more than 0.02 and the duration of more than 5 seconds is regarded as an uphill working condition, data under the uphill working condition is reserved, data under other working conditions are removed, and a road resistance coefficient expression is shown as follows:
wherein F is t ＝T e i g i 0 η t R is the driving force，F w ＝C d Av 2 Air resistance/21.15, F j ＝δm ass dv/dt is acceleration resistance, m ass Representing the mass of the vehicle, g represents the gravitational acceleration, i g Representing the transmission ratio of each gear, i 0 Represents the main reduction gear ratio, r represents the radius of the wheel, eta t Representing transmission efficiency, C d Represents the air resistance coefficient, A represents the windward area of the vehicle, delta represents the conversion coefficient of the rotating mass and T e Represents engine torque, and v represents vehicle speed.
The construction of the driving data set and the removal of outliers described in step 4) is performed as follows:
step 4-1), firstly, selecting a gear shifting control parameter under a target working condition, taking an ascending working condition as an example, wherein factors influencing the gear shifting of a vehicle under the ascending working condition mainly comprise a vehicle speed, an accelerator pedal opening, acceleration and gradient; the downhill working condition can be the same parameter as the uphill working condition, and the road resistance coefficient is a negative value.
Step 4-2) extracting a data set in each gear composed of the speed, the acceleration and the accelerator pedal opening under the acceleration working condition from the uphill working condition data obtained after pretreatment and data cleaning;
and 4-3) carrying out pretreatment and data cleaning to obtain data points in scattered distribution at the outermost edge part of the data set, wherein the data points are caused by abnormal operation of a driver, and detecting and processing the outliers in the data set by using a support vector machine algorithm (One Class SVM, OCSVM) algorithm. The OCSVM algorithm has excellent outlier processing effect, so that the OCSVM algorithm is selected to detect and process outliers in the data set. The OCSVM algorithm is an unsupervised learning algorithm whose basic idea is to calculate a minimum radius hypersphere in the dataset and to include all test samples inside this hypersphere. Let the center of the super sphere be o, the radius of the corresponding super sphere be R, and ensure the super sphereThe volume radius is as small as possible and contains as many sample points as possible. Let the training sample set be t= { X i |i＝1,2,…,m,X i ∈R M Constructing a relaxation variable ζ with a penalty coefficient C i The optimization problem is expressed as:
s.t.|φ(X i )-o|| 2 ≤R+ζ i ,ζ i ≥0,i＝1,2,…,m
where m represents the number of training sample points,
representing a nonlinear mapping of the original M-dimensional space to some high-dimensional feature space. The dual form of the above can be obtained by using the Lagrangian multiplier method and introducing a kernel function as follows:
wherein alpha is i，j Represents the Lagrangian multiplier, K (X i ,X j ) Representing the kernel function, a decision function can be constructed for a new sample z, other than the training sample, that needs to be judged whether it is an outlier:
wherein the kernel function is selected as a Gaussian radial basis kernel function, and whether the test data point is in the hypersphere can be judged through a decision function, and if g (z)>R 2 The test data point is judged to be an outlier.
And step 6), fitting the gear shifting boundary points extracted in the step 5) by using a moving least square method so as to obtain a gear shifting regular curved surface. The moving least square method improves the least square method, and can overcome the difficulties of block fitting and smoothing in the traditional least square method in the surface fitting. The least squares method is to form a fitting function f (x) approximation function by a base function p (x) and a coefficient vector a (x), and the fitting function f (x) approximation function is shown in the following formula:
wherein f h (x) Is an approximation function of the fitting function f (x). x { x, y } represents the bounded domain Ω in space R 2 In (a) = [ a ], α (x) = [ a ] 1 (x),a 2 (x),Λ,a m (x)] T Is a coefficient to be determined. p (x) = [ p ] 1 (x),p 2 (x),Λ,p m (x)] T As basis functions, the basis functions of the present invention select complete quadratic basis functions in two-dimensional space:
p(x)＝(1,x,y,x 2 ,xy,y 2 ) T
in order to obtain an accurate local approximation, let the discrete weighted paradigm of the residual be:
where n is the number of nodes in the solution area, w (x-x i ) Is the weight function of the node, x i Representative node (i=1, 2, ﹍, n), y i ＝f(x i ). Formula (14) is rewritten into a matrix form:
J＝[pa-f] T W[pa-f]
wherein, the liquid crystal display device comprises a liquid crystal display device,
a＝A -1 (x)B(x)f(x)
A(x)＝p T W(x)p
B(x)＝p T W(x)
where l represents the number of terms of the basis function, where l=6.
From this, an approximation function can be obtained as:
f h (x)＝p T (x)A -1 (x)B(x)f(x)
the choice of the weight function is very important for the choice of the mobile least squares method, the weight function is non-negative and as the fitting point is distant from the node, ||x-x i The increase in i decreases. The weight function should also have a compactness of greater than 0 within the support domain and 0 outside the support domain. The invention selects a cubic spline weight function as a fitted weight function, so that s=x-x i ，
The cubic spline weight function is: />
And fitting a gear shifting boundary point by using a moving least square method, so as to obtain a gear shifting strategy of an excellent driver for operating the vehicle under an uphill working condition.
The invention is further described below with reference to the accompanying drawings.
In the gear decision method based on data mining, massive driving data of a manual gear vehicle driven by an excellent driver in various road environments are required to be collected, the collected driving conditions mainly comprise urban roads, suburban roads, mountain roads and urban expressways, the driving route of data collection is shown in fig. 1, and the flow of gear decision construction by using a processed data set is shown in fig. 2;
the acquired driving data needs to be subjected to data preprocessing, and the flow of the data preprocessing is shown in fig. 3, and the data preprocessing comprises data synchronization, data resampling, missing value processing and data filtering processing, wherein a data synchronization schematic diagram is shown in fig. 4, and a data resampling schematic diagram is shown in fig. 5 (taking an acceleration signal as an example).
The collected driving data is subjected to data preprocessing, and then the driving data under the target working condition is extracted through a series of data cleaning, wherein the flow of data cleaning is shown in fig. 6 and comprises data segment division, curve data cleaning and ramp data extraction.
After excellent driver driving data under a target working condition is obtained, proper gear decision parameters are selected according to the characteristics of the working condition, and the intelligent gear decision method based on data mining is introduced by the uphill working condition. The uphill working condition is further divided into an acceleration uphill working condition and a uniform speed uphill working condition, wherein gear decision parameters under the acceleration uphill working condition are selected as a vehicle speed, an accelerator pedal opening and an acceleration, and gear decision parameters under the uniform speed uphill working condition are selected as the vehicle speed, the accelerator pedal opening and a road resistance coefficient.
The data set shown in fig. 7 can be constructed from target working condition data obtained after data cleaning by using gear decision parameters, data points scattered at the outermost edge part of the data set are caused by abnormal operation of a driver, outliers in the data set are detected and processed by using a support vector machine algorithm, and the data set after the outliers are processed is shown in fig. 7.
The left side and the right side of the data set under each gear along the speed increasing direction are respectively used as boundaries of the downshift and the upshift, so that boundary points of data of each gear are required to be extracted.
The gear-up regular curved surface and the gear-down regular curved surface of each gear can be obtained by fitting the extracted gear-shifting boundary points by using a moving least square method, and the gear-up regular curved surface obtained by fitting is shown in fig. 9.
Finally, it should be noted that: the above description is only illustrative of the specific embodiments of the invention and it is of course possible for those skilled in the art to make modifications and variations to the invention, which are deemed to be within the scope of the invention as defined in the claims and their equivalents.
Claims (8)
1. The intelligent gear decision design method of the automatic transmission based on data mining is characterized by comprising the following steps of:
step S1: acquiring driving data of a good driver driving a manual transmission vehicle in various road environments;
step S2: preprocessing the acquired driving data, including data synchronization, data resampling, data filtering and missing value processing;
step S3: the method comprises the steps of performing data cleaning on the preprocessed driving data, extracting driving data under a target working condition, constructing gear decision sub-working conditions, and obtaining a complete excellent driver gear decision strategy by dividing various driving working conditions and respectively constructing gear decision strategies;
step S4: three gear shifting features under a target working condition are selected, a driving data set is constructed, outliers in the data set are detected by using a support vector machine algorithm, and the outliers in the data set are removed;
step S5: extracting a gear shifting boundary point of each gear data in the data set by using an Alpha-shapes algorithm;
step S6: and fitting the extracted gear shifting boundary points by using a mobile least square method to obtain a gear shifting regular curved surface of each gear.
2. The data mining-based intelligent gear decision design method for the automatic transmission, according to claim 1, is characterized in that: in the step S2, the data resampling includes selecting a reference frequency, and resampling data of other frequencies by the reference frequency to ensure the unification of sampling frequencies; carrying out synchronization and resampling processing on the data by adopting a cubic spline curve interpolation algorithm; filtering noise of an original driving data set by adopting a Butterworth filter; defining data with zero speed and duration exceeding T seconds in the original driving data set as a missing value, wherein T is more than or equal to 2; and processing the missing value by adopting a method for deleting the original driving data in the time period of the missing value.
3. The data mining-based intelligent gear decision design method for the automatic transmission, according to claim 2, is characterized in that: in the step S2, the preprocessing of the driving data of the excellent driver is performed according to the following steps:
step S21: exporting the acquired driving data files into excel files, wherein the size of each excel file is not more than 120MB, determining the time starting point of each signal in each excel file, if the time starting points are inconsistent, selecting the time with the maximum time starting point as a unified time starting point, and taking the minimum ending time point of each signal as a unified ending point;
step S22: after determining a starting time point and an ending time point in each excel file, interpolating by using a cubic spline interpolation algorithm at intervals of 0.01s so as to obtain running data with the sampling frequency of 100Hz, thereby completing data synchronization and resampling in data preprocessing;
step S23: the frequency response of the butterworth filter is shown as follows:
in the formula (1), N is the order of the filter, ω represents the circular frequency, ω p For passband edge frequency omega c For a 3dB passband cut-off frequency, ε represents the normalized tuning parameter, satisfying:
wherein alpha is p Is the passband maximum attenuation coefficient; will |H (j omega) | 2 The form rewritten as a laplace function is:
wherein, the expression of the transfer function H(s) is:
wherein s is k Representing the pole, let p=s/ω c The above equation is normalized with a 3dB cut-off frequency, which is converted to a transfer function H (p):
wherein p is k Represents a normalized pole; by passband edge frequency omega p Stop band cut-off frequency omega s Maximum attenuation coefficient alpha of stop band s The filter order may be determined:
after determining the order of the filter, obtaining a normalized pole, and then obtaining H(s) through de-normalization, thereby determining the transfer function of the filter.
4. The data mining-based intelligent gear decision design method for the automatic transmission, according to claim 1, is characterized in that: in step S3, performing data cleaning on the preprocessed driving data to extract driving data under the target working condition includes:
and after the missing values are removed, taking the effective data between two adjacent missing value data as a data segment, and carrying out subsequent data cleaning by judging the working condition corresponding to each data segment.
5. The data mining-based intelligent gear decision design method for the automatic transmission, as set forth in claim 4, is characterized in that: the working conditions corresponding to the data segments are uphill working conditions, and curve data, bumpy road data, low-adhesion road surface and downhill road data in the driving data need to be removed in the data cleaning process.
6. The data mining-based intelligent gear decision design method for the automatic transmission, as set forth in claim 5, is characterized in that: firstly, taking effective data between two adjacent missing value data as a data segment, and judging the vehicle running condition corresponding to each data segment by utilizing the steering wheel angle and the road resistance coefficient so as to carry out subsequent data cleaning;
in the curve data cleaning process, firstly calculating the steering wheel angle absolute value of each data segment, and then regarding the data segment with the steering wheel angle absolute value being more than 35 degrees and the duration exceeding 3 seconds as curve data and removing the curve data;
the change of the road resistance coefficient is due to the change of the vehicle weight, the road gradient and the rolling resistance, the data of the road resistance coefficient larger than the rolling resistance coefficient is regarded as the ascending working condition, and the road resistance coefficient f is considered in consideration of the fluctuation of the road resistance coefficient road Data with a duration greater than 5 seconds and greater than 0.02 is considered an uphill condition, data is retained during the uphill condition, and data is removed during other conditions.
7. The data mining-based intelligent gear decision design method for the automatic transmission, according to claim 1, is characterized in that: the construction of the driving data set and the removal of outliers described in the step 4 are performed as follows:
step 4-1: firstly, selecting a gear shifting control parameter under a target working condition, wherein factors influencing gear shifting of a vehicle under an ascending working condition include a vehicle speed, an accelerator pedal opening, acceleration and gradient, the ascending working condition is further divided into a uniform ascending working condition and an accelerating ascending working condition, the uniform ascending working condition takes the vehicle speed, the accelerator pedal opening and a road resistance coefficient as the gear shifting parameter, and the accelerating ascending working condition takes the vehicle speed, the accelerator pedal opening and the acceleration as the gear shifting parameter;
step 4-2: extracting a data set of each gear composed of the speed, the acceleration and the accelerator pedal opening under the acceleration working condition from the uphill working condition data obtained after pretreatment and data cleaning;
step 4-3: and detecting and processing the outliers in the data set by using a support vector machine algorithm.
8. The data mining-based intelligent gear decision design method for the automatic transmission, according to claim 1, is characterized in that: in step S5, a gear shift boundary point of each gear data in the dataset is extracted by using an Alpha-shapes algorithm, a vehicle speed maximum side boundary point of each gear data along the vehicle speed increasing direction is used as an upshift boundary point, and a vehicle speed minimum side boundary point is used as a downshift boundary point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210882192.8A CN115264048B (en) | 2022-07-26 | 2022-07-26 | Intelligent gear decision design method for automatic transmission based on data mining |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210882192.8A CN115264048B (en) | 2022-07-26 | 2022-07-26 | Intelligent gear decision design method for automatic transmission based on data mining |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115264048A CN115264048A (en) | 2022-11-01 |
| CN115264048B true CN115264048B (en) | 2023-05-23 |
Family
ID=83768588
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210882192.8A Active CN115264048B (en) | 2022-07-26 | 2022-07-26 | Intelligent gear decision design method for automatic transmission based on data mining |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115264048B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116906556B (en) * | 2023-07-11 | 2024-04-09 | 徐州徐工传动科技有限公司 | Intelligent gear decision method for automatic transmission turning lane changing working condition |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180093671A1 (en) * | 2017-11-21 | 2018-04-05 | GM Global Technology Operations LLC | Systems and methods for adjusting speed for an upcoming lane change in autonomous vehicles |
| CN110502552B (en) * | 2019-08-20 | 2022-10-28 | 重庆大学 | Classification data conversion method based on fine-tuning conditional probability |
| CN110682914A (en) * | 2019-09-29 | 2020-01-14 | 皖西学院 | Driving behavior recognition system and method based on wireless perception |
| CN111717217B (en) * | 2020-06-30 | 2022-11-08 | 重庆大学 | Driver intention identification method based on probability correction |
| CN114412988B (en) * | 2022-02-11 | 2022-11-29 | 吉林大学 | Uphill predictive gear shifting control method based on road information and machine learning |
-
2022
- 2022-07-26 CN CN202210882192.8A patent/CN115264048B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115264048A (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109131350B (en) | Energy management method and system for hybrid electric vehicle | |
| CN110991757B (en) | Comprehensive prediction energy management method for hybrid electric vehicle | |
| CN109733248A (en) | Pure electric automobile remaining mileage model prediction method based on routing information | |
| CN115264048B (en) | Intelligent gear decision design method for automatic transmission based on data mining | |
| CN110985651B (en) | Automatic transmission multi-parameter fusion gear shifting strategy based on prediction | |
| US20150066319A1 (en) | Method for controlling shift of automatic transmission in vehicle | |
| CN113525396B (en) | Hybrid electric vehicle layered prediction energy management method integrating deep reinforcement learning | |
| CN109555847B (en) | Hybrid power bus AMT gear shifting method based on dynamic programming | |
| Vahidi et al. | Experiments for online estimation of heavy vehicle’s mass and time-varying road grade | |
| Miao et al. | Three-parameter transmission gear-shifting schedule for improved fuel economy | |
| CN112429005A (en) | Pure electric vehicle personalized gear shifting rule optimization method considering transmission efficiency and application | |
| CN112677982A (en) | Vehicle longitudinal speed planning method based on driver characteristics | |
| CN103398168A (en) | Modeling method for gear-shift schedule of automatic transmission | |
| CN113159435B (en) | Method and system for predicting remaining driving mileage of new energy vehicle | |
| CN113401123B (en) | Automobile prediction cruise parameter self-tuning control system fusing driving mode information | |
| US5794169A (en) | System for determining the shift stage of an automatic transmission by using fuzzy inference to decide road gradient | |
| CN115492928B (en) | Economical efficiency, dynamic performance and safety comprehensive optimal gear shifting rule optimization method | |
| CN104653760B (en) | Gear shifting control method and gear shifting control device for AMT (Automated Mechanical Transmission) and AMT | |
| CN117109941A (en) | Vehicle testing method and device, electronic equipment and storage medium | |
| CN114255596B (en) | Parking lot parking space recommendation system and method based on big data | |
| Sagoian et al. | Energy consumption prediction of electric vehicle air conditioning system using artificial intelligence | |
| CN112489422B (en) | Traffic flow prediction method based on algorithm model analysis | |
| Cheng et al. | Intelligent gear decision method for automatic vehicles based on data mining under uphill conditions | |
| CN116906556B (en) | Intelligent gear decision method for automatic transmission turning lane changing working condition | |
| CN112528568A (en) | Road feel simulation method based on K-Means and BP neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |