CN114407847A - Single-pedal wire-controlled chassis automobile auxiliary braking method based on machine learning - Google Patents
Single-pedal wire-controlled chassis automobile auxiliary braking method based on machine learning Download PDFInfo
- Publication number
- CN114407847A CN114407847A CN202210179569.3A CN202210179569A CN114407847A CN 114407847 A CN114407847 A CN 114407847A CN 202210179569 A CN202210179569 A CN 202210179569A CN 114407847 A CN114407847 A CN 114407847A
- Authority
- CN
- China
- Prior art keywords
- steering wheel
- maximum value
- longitudinal control
- pedal
- control pedal
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000010801 machine learning Methods 0.000 title claims abstract description 13
- 230000001133 acceleration Effects 0.000 claims abstract description 79
- 230000006399 behavior Effects 0.000 claims abstract description 8
- 239000002245 particle Substances 0.000 claims abstract description 8
- 230000003044 adaptive effect Effects 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 17
- 238000012544 monitoring process Methods 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 4
- 230000001172 regenerating effect Effects 0.000 claims description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 claims description 3
- 230000001788 irregular Effects 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000012886 linear function Methods 0.000 claims 1
- 206010039203 Road traffic accident Diseases 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000003211 malignant effect Effects 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60T—VEHICLE BRAKE CONTROL SYSTEMS OR PARTS THEREOF; BRAKE CONTROL SYSTEMS OR PARTS THEREOF, IN GENERAL; ARRANGEMENT OF BRAKING ELEMENTS ON VEHICLES IN GENERAL; PORTABLE DEVICES FOR PREVENTING UNWANTED MOVEMENT OF VEHICLES; VEHICLE MODIFICATIONS TO FACILITATE COOLING OF BRAKES
- B60T7/00—Brake-action initiating means
- B60T7/12—Brake-action initiating means for automatic initiation; for initiation not subject to will of driver or passenger
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60K—ARRANGEMENT OR MOUNTING OF PROPULSION UNITS OR OF TRANSMISSIONS IN VEHICLES; ARRANGEMENT OR MOUNTING OF PLURAL DIVERSE PRIME-MOVERS IN VEHICLES; AUXILIARY DRIVES FOR VEHICLES; INSTRUMENTATION OR DASHBOARDS FOR VEHICLES; ARRANGEMENTS IN CONNECTION WITH COOLING, AIR INTAKE, GAS EXHAUST OR FUEL SUPPLY OF PROPULSION UNITS IN VEHICLES
- B60K26/00—Arrangements or mounting of propulsion unit control devices in vehicles
- B60K26/02—Arrangements or mounting of propulsion unit control devices in vehicles of initiating means or elements
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60T—VEHICLE BRAKE CONTROL SYSTEMS OR PARTS THEREOF; BRAKE CONTROL SYSTEMS OR PARTS THEREOF, IN GENERAL; ARRANGEMENT OF BRAKING ELEMENTS ON VEHICLES IN GENERAL; PORTABLE DEVICES FOR PREVENTING UNWANTED MOVEMENT OF VEHICLES; VEHICLE MODIFICATIONS TO FACILITATE COOLING OF BRAKES
- B60T7/00—Brake-action initiating means
- B60T7/02—Brake-action initiating means for personal initiation
- B60T7/04—Brake-action initiating means for personal initiation foot actuated
- B60T7/06—Disposition of pedal
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/72—Electric energy management in electromobility
Landscapes
- Engineering & Computer Science (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Control Of Driving Devices And Active Controlling Of Vehicle (AREA)
Abstract
The invention discloses a machine learning-based auxiliary braking method for a single-pedal wire-controlled chassis automobile, which comprises the following steps of: sample data acquisition; constructing a driver behavior model based on a support vector regression algorithm and an adaptive particle swarm algorithm; whether a driver operates a longitudinal control pedal in real time or not is monitored, after the system judges that the driver operates the longitudinal control pedal in a wrong way, a single-pedal driving mode of the automobile is changed into a braking mode, when the driver steps on the longitudinal control pedal downwards, the automobile receives a braking signal, a braking executing mechanism starts executing, and meanwhile, an acceleration signal is cut off, so that the safety of the automobile is ensured, and traffic accidents caused by unexpected rapid acceleration are avoided.
Description
Technical Field
The invention relates to the technical field of automobiles, in particular to a single-pedal wire-control chassis automobile auxiliary braking method based on machine learning.
Background
With the development of automotive technology, chassis-by-wire automobiles with a single-pedal driving mode have appeared, and a longitudinal control pedal is adopted to replace a conventional accelerator pedal. The longitudinal control pedal has the characteristic of acceleration and braking integration, and the working principle is that when a driver steps on the longitudinal control pedal, the motor outputs driving torque to realize the acceleration of the automobile; after the driver looses the longitudinal control pedal, the motor outputs braking torque to realize the deceleration of the automobile through regenerative braking.
However, after the driver has adapted to the settings of accelerator pedal depression and brake pedal release, when emergency braking is required due to an emergency, the presence of the brake pedal is likely to be ignored and the longitudinal control pedal is likely to be misoperated due to driving habits. At present, many single-pedal automobile accidental acceleration accidents occur at home and abroad, and the accidents are all caused by improper use of the pedal by a driver according to investigation. How to reduce the occurrence of such malignant accidents to the maximum extent and further improve the safety of the automobile is very reluctant. It is necessary to research a single-pedal wire-controlled chassis automobile auxiliary braking method which can accurately sense the dangerous behavior of the misoperation of the pedal of the driver and help the driver to take braking measures so as to avoid or reduce casualty accidents.
Disclosure of Invention
The invention aims to provide a single-pedal drive-by-wire chassis automobile auxiliary braking method based on machine learning to solve the technical problem.
The auxiliary braking method for the single-pedal wire-controlled chassis automobile comprises the following steps:
s1: sample data acquisition:
selecting 100 male testers and 100 female testers with the ages of C2 and above and the ages of the male testers and the female testers which are evenly distributed between 18 and 60 years old to carry out a real vehicle test in a driving field, wherein the real vehicle adopts an acceleration and braking integrated longitudinal control pedal to replace a traditional acceleration pedal, a driver steps on the longitudinal control pedal to accelerate, the longitudinal control pedal is released to decelerate through regenerative braking, and a brake pedal consistent with the traditional vehicle is arranged on the left side of the longitudinal control pedal and used for emergency braking; the method comprises the steps of collecting information of age, driving age, height and weight of each tester before a test is started, and then testing each tester under three working conditions of stable acceleration, rapid acceleration and longitudinal control pedal misoperation when encountering an obstacle, wherein 10 times of testing are arranged in each working condition, and the three working conditions randomly appear in the test process; the method comprises the steps of monitoring the speed of the automobile and the distance between the automobile and a front obstacle in real time in the test process, and collecting the maximum value of the angular acceleration of a longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of a steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel of all testers under the working conditions that the testers suddenly accelerate and the longitudinal control pedal is operated by the obstacles in a misoperation mode.
S2: constructing a driver behavior model:
based on the test data, an end-to-end prediction model of the age, the driving age, the height, the weight, the speed and the distance between the automobile and the front obstacle, which is related to the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the rotation angle of the steering wheel, the maximum value of the rotation angular velocity of the steering wheel and the maximum value of the rotation angular acceleration of the steering wheel, is established, and the end-to-end prediction model is established through a support vector regression model.
And taking the age, driving age, height, weight, speed and distance between the automobile and a front obstacle in the sample data as input variables of the support vector regression model, and taking the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel in the sample data as output variables of the support vector regression model. And sending the training data set after normalization processing into a support vector regression model for training, predicting the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the corner of the steering wheel, the maximum value of the rotating angular speed of the steering wheel and the maximum value of the rotating angular acceleration of the steering wheel by using the learned high-dimensional mapping relation, and providing a data basis for establishing a discrimination logic that a driver mistakenly steps on the pedal to realize mistaken acceleration.
The method comprises the following specific steps:
carrying out normalization processing on the sample data according to the following equation:
wherein, XminIs the minimum value of the sample data; xmaxIs the maximum value of the sample data; x is sample data; x' is normalized data in the range of 0,1]。
After the sample data is normalized, dividing the normalized data into two parts, wherein 80% of the normalized data is classified as a training data set, and 20% of the normalized data is classified as a testing data set.
Integrating the sample data into a data set D { (x)1,y1),(x2,y2)...,(xn,yn)}，
The training samples are mapped from a low-dimensional space to a high-dimensional space through nonlinear mapping, and a linear regression model established in the high-dimensional space can be expressed as the following equation:
f(x)＝w·Φ(x)+b
where x is the input variable, Φ (x) is the nonlinear function that maps x to a high-dimensional linear space, w is the weight vector, and b is the offset.
In order to minimize the regression error, the objective function of the vector regression algorithm is supported, and the following equation can be expressed:
wherein, CpIs a penalty coefficient which represents the penalty degree of the model to the sample with the error larger than epsilon in the training process, lεIs an epsilon-insensitive loss function, epsilon represents an insensitive loss coefficient, the smaller epsilon represents the smaller error requirement of the regression function, and lεThe expression can be expressed as the following equation:
wherein z represents the error between the fitting value and the true value of the support vector regression algorithm.
When the data does not conform to lε(z) constraint, introducing a relaxation variable deltai，δi *To correct for the irregular factor, the following equation can then be derived:
by introducing lagrange multiplier alphai、αi *Simplifying the calculation, converting the above formula into the solution of alphai，αi *The dual problem of (2):
wherein, K (x)i,xj) The method selects an RBF kernel function, and the RBF kernel function is defined as the following equation:
K(xi,xj)＝exp(-γ||xi-xj||2)
wherein γ is a nuclear parameter.
According to the carlo-kun-tower condition, the regression function f (x) obtained by solving can be expressed as:
based on the above method, the driver behavior model can be abstracted as:
y＝f(x|(Cp,ε,γ))。
then, using the adaptive particle swarm optimization algorithm to support three hyper-parameters of the vector regression model, namely penalty coefficient CpAnd optimizing a kernel parameter gamma and an insensitive loss coefficient epsilon. Selecting an average absolute percentage error MAPE capable of directly reflecting regression performance as a fitness function fitness of the self-adaptive particle swarm algorithm, namely:
where n is the number of sample data, yiIs the predicted value, f (x)i) Are experimental values.
Predicting the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the rotation angle of the steering wheel, the maximum value of the rotation angular velocity of the steering wheel and the maximum value of the rotation angular acceleration of the steering wheel by using a support vector regression model obtained by training, and adopting a Mean Square Error (MSE) and a decision coefficient R2And evaluating the prediction result of the model:
wherein,
is the average of the predicted values and is,
is the average of the experimental values.
S3: monitoring whether the driver operates the longitudinal control pedal in a misoperation mode in real time:
collecting the age, the driving age, the height and the weight of a driver before the driver starts to drive the automobile, monitoring the speed of the automobile and the distance between the automobile and a front obstacle in real time in the driving process of the automobile, and monitoring the angular acceleration a of a longitudinal control pedal, the pressure h of the longitudinal control pedal, the hand holding force f of a steering wheel, the torque m of the steering wheel, the turning angle theta of the steering wheel, the rotational angular speed omega of the steering wheel and the rotational angular acceleration beta of the steering wheel; meanwhile, the maximum value a of the angular acceleration of the longitudinal control pedal of the driver under the condition of rapid acceleration is calculated through a support vector regression model1Longitudinal control of maximum pedal pressure h1Maximum hand holding force f of steering wheel1Steering wheel torque maximum m1Maximum steering wheel angle θ1Maximum value of angular velocity ω of steering wheel rotation1And maximum value of angular acceleration of steering wheel1And maximum angular acceleration a of the longitudinal control pedal under the condition that the longitudinal control pedal is operated by mistake in case of an obstacle2Longitudinal control of maximum pedal pressure h2Maximum hand holding force f of steering wheel2Steering wheel torque maximum m2Maximum steering wheel angle θ2Maximum value of angular velocity ω of steering wheel rotation2And maximum value of angular acceleration of steering wheel2。
When any one of the following two conditions occurs, the system judges that the driver operates the longitudinal control pedal by mistake and starts the auxiliary brake module:
the first condition is as follows: the following seven conditions simultaneously satisfy four and more:
①a1≤a＜a2；
②h1≤h＜h2；
③f1≤f＜f2；
④m1≤m＜m2；
⑤θ1≤θ＜θ2；
⑥ω1≤ω＜ω2；
⑦β1≤β＜β2；
case two: any one of the following seven conditions is satisfied:
①a≥a2；
②h≥h2；
③f≥f2；
④m≥m2；
⑤θ≥θ2；
⑥ω≥ω2；
⑦β≥β2。
the auxiliary brake module starts to work after the system judges that a driver mistakenly operates a longitudinal control pedal, a single-pedal driving mode of an automobile integrating acceleration and braking is changed into a brake mode, when the driver downwards steps on the longitudinal control pedal, the automobile receives a brake signal, a brake executing mechanism starts to execute, the acceleration signal is cut off at the same time, when the pressure and the speed of the longitudinal control pedal are both 0, the auxiliary brake module is recovered, the automobile is recovered to be in the single-pedal driving mode, and then the driver can normally drive the automobile, steps on the longitudinal control pedal to accelerate the automobile and releases the longitudinal control pedal to decelerate the automobile.
The invention has the beneficial effects that:
1. based on the adaptive particle swarm algorithm and the support vector regression algorithm, a proxy model of the age, the driving age, the height, the weight, the speed of the vehicle and the distance between the vehicle and a front obstacle, which is related to the maximum acceleration value of an accelerator pedal, the maximum pressure value of the accelerator pedal, the maximum hand holding force of a steering wheel, the maximum torque value of the steering wheel, the maximum turning angle of the steering wheel, the maximum turning angular velocity of the steering wheel and the maximum turning angular acceleration of the steering wheel, is established.
2. After the system judges that the driver mistakenly operates the longitudinal control pedal, the automobile is changed from a single-pedal driving mode to a braking mode, when the driver downwards steps on the longitudinal control pedal, the automobile receives a braking signal, the braking execution mechanism starts executing the braking signal, and the acceleration signal is cut off, so that traffic accidents caused by unexpected rapid acceleration are avoided.
3. The method has strong universality and is generally suitable for automobiles produced by different manufacturers.
Drawings
In order to more clearly illustrate the detailed description of the invention or the technical solutions in the prior art, the drawings that are needed in the detailed description of the invention or the prior art will be briefly described below.
FIG. 1 is a system framework diagram of a single-pedal drive-by-wire chassis automobile auxiliary braking method based on machine learning according to the present invention.
Detailed Description
The following embodiments are only used for illustrating the technical solutions of the present invention more clearly, and therefore, the following embodiments are only used as examples, and the protection scope of the present invention is not limited thereby.
It is to be noted that, unless otherwise specified, technical or scientific terms used herein shall have the ordinary meaning as understood by those skilled in the art to which the invention pertains.
A machine learning-based auxiliary braking method for a single-pedal drive-by-wire chassis automobile is shown in a system frame diagram of fig. 1.
The auxiliary braking method for the single-pedal wire-controlled chassis automobile comprises the following steps:
the auxiliary braking method for the single-pedal wire-controlled chassis automobile comprises the following steps:
s1: sample data acquisition:
selecting 100 male testers and 100 female testers with the ages of C2 and above and the ages of the male testers and the female testers which are evenly distributed between 18 and 60 years old to carry out a real vehicle test in a driving field, wherein the real vehicle adopts an acceleration and braking integrated longitudinal control pedal to replace a traditional acceleration pedal, a driver steps on the longitudinal control pedal to accelerate, the longitudinal control pedal is released to decelerate through regenerative braking, and a brake pedal consistent with the traditional vehicle is arranged on the left side of the longitudinal control pedal and used for emergency braking; the method comprises the steps of collecting information of age, driving age, height and weight of each tester before a test is started, and then testing each tester under three working conditions of stable acceleration, rapid acceleration and longitudinal control pedal misoperation when encountering an obstacle, wherein 10 times of testing are arranged in each working condition, and the three working conditions randomly appear in the test process; the method comprises the steps of monitoring the speed of the automobile and the distance between the automobile and a front obstacle in real time in the test process, and collecting the maximum value of the angular acceleration of a longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of a steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel of all testers under the working conditions that the testers suddenly accelerate and the longitudinal control pedal is operated by the obstacles in a misoperation mode.
S2: constructing a driver behavior model:
based on the test data, an end-to-end prediction model of the age, the driving age, the height, the weight, the speed and the distance between the automobile and the front obstacle, which is related to the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the rotation angle of the steering wheel, the maximum value of the rotation angular velocity of the steering wheel and the maximum value of the rotation angular acceleration of the steering wheel, is established, and the end-to-end prediction model is established through a support vector regression model.
And taking the age, driving age, height, weight, speed and distance between the automobile and a front obstacle in the sample data as input variables of the support vector regression model, and taking the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel in the sample data as output variables of the support vector regression model. And sending the training data set after normalization processing into a support vector regression model for training, predicting the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the corner of the steering wheel, the maximum value of the rotating angular speed of the steering wheel and the maximum value of the rotating angular acceleration of the steering wheel by using the learned high-dimensional mapping relation, and providing a data basis for establishing a discrimination logic that a driver mistakenly steps on the pedal to realize mistaken acceleration.
The method comprises the following specific steps:
carrying out normalization processing on the sample data according to the following equation:
wherein, XminIs the minimum value of the sample data; xmaxIs the maximum value of the sample data; x is sample data; x' is normalized data in the range of 0,1]。
After the sample data is normalized, dividing the normalized data into two parts, wherein 80% of the normalized data is classified as a training data set, and 20% of the normalized data is classified as a testing data set.
Integrating the sample data into a data set D { (x)1,y1),(x2,y2)...,(xn,yn)}，
The training samples are mapped from a low-dimensional space to a high-dimensional space through nonlinear mapping, and a linear regression model established in the high-dimensional space can be expressed as the following equation:
f(x)＝w·Φ(x)+b
where x is the input variable, Φ (x) is the nonlinear function that maps x to a high-dimensional linear space, w is the weight vector, and b is the offset.
In order to minimize the regression error, the objective function of the vector regression algorithm is supported, and the following equation can be expressed:
wherein, CpIs a penalty coefficient which represents the penalty degree of the model to the sample with the error larger than epsilon in the training process, lεIs an epsilon-insensitive loss function, epsilon represents an insensitive loss coefficient, the smaller epsilon represents the smaller error requirement of the regression function, and lεThe expression can be expressed as the following equation:
wherein z represents the error between the fitting value and the true value of the support vector regression algorithm.
When the data does not conform to lε(z) constraint, introducing a relaxation variable deltai，δi *To correct for the irregular factor, the following equation can then be derived:
by introducing lagrange multiplier alphai、αi *Simplifying the calculation, converting the above formula into the solution of alphai，αi *The dual problem of (2):
wherein, K (x)i,xj) The method selects an RBF kernel function, and the RBF kernel function is defined as the following equation:
K(xi,xj)＝exp(-γ||xi-xj||2)
wherein γ is a nuclear parameter.
According to the carlo-kun-tower condition, the regression function f (x) obtained by solving can be expressed as:
based on the above method, the driver behavior model can be abstracted as:
y＝f(x|(Cp,ε,γ))。
then, using the adaptive particle swarm optimization algorithm to support three hyper-parameters of the vector regression model, namely penalty coefficient CpAnd optimizing a kernel parameter gamma and an insensitive loss coefficient epsilon. Selecting an average absolute percentage error MAPE capable of directly reflecting regression performance as a fitness function fitness of the self-adaptive particle swarm algorithm, namely:
where n is the number of sample data, yiIs the predicted value, f (x)i) Are experimental values.
Predicting the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the rotation angle of the steering wheel, the maximum value of the rotation angular velocity of the steering wheel and the maximum value of the rotation angular acceleration of the steering wheel by using a support vector regression model obtained by training, and adopting a Mean Square Error (MSE) and a decision coefficient R2And evaluating the prediction result of the model:
wherein,
is the average of the predicted values and is,
is the average of the experimental values.
S3: monitoring whether the driver operates the longitudinal control pedal in a misoperation mode in real time:
collecting the age, the driving age, the height and the weight of a driver before the driver starts to drive the automobile, monitoring the speed of the automobile and the distance between the automobile and a front obstacle in real time in the driving process of the automobile, and monitoring the angular acceleration a of a longitudinal control pedal, the pressure h of the longitudinal control pedal, the hand holding force f of a steering wheel, the torque m of the steering wheel, the turning angle theta of the steering wheel, the rotational angular speed omega of the steering wheel and the rotational angular acceleration beta of the steering wheel; meanwhile, the maximum value a of the angular acceleration of the longitudinal control pedal of the driver under the condition of rapid acceleration is calculated through a support vector regression model1Longitudinal control of maximum pedal pressure h1Maximum hand holding force f of steering wheel1Steering wheel torque maximum m1Maximum steering wheel angle θ1Maximum value of angular velocity ω of steering wheel rotation1And maximum value of angular acceleration of steering wheel1And maximum angular acceleration a of the longitudinal control pedal under the condition that the longitudinal control pedal is operated by mistake in case of an obstacle2Longitudinal control of maximum pedal pressure h2Maximum hand holding force f of steering wheel2Steering wheel torque maximum m2Maximum steering wheel angle θ2Maximum value of angular velocity ω of steering wheel rotation2And maximum value of angular acceleration of steering wheel2。
When any one of the following two conditions occurs, the system judges that the driver operates the longitudinal control pedal by mistake and starts the auxiliary brake module:
the first condition is as follows: the following seven conditions simultaneously satisfy four and more:
①a1≤a＜a2；
②h1≤h＜h2；
③f1≤f＜f2；
④m1≤m＜m2；
⑤θ1≤θ＜θ2；
⑥ω1≤ω＜ω2；
⑦β1≤β＜β2；
case two: any one of the following seven conditions is satisfied:
①a≥a2；
②h≥h2；
③f≥f2；
④m≥m2；
⑤θ≥θ2；
⑥ω≥ω2；
⑦β≥β2。
the auxiliary brake module starts to work after the system judges that a driver mistakenly operates a longitudinal control pedal, a single-pedal driving mode of an automobile integrating acceleration and braking is changed into a brake mode, when the driver downwards steps on the longitudinal control pedal, the automobile receives a brake signal, a brake executing mechanism starts to execute, the acceleration signal is cut off at the same time, when the pressure and the speed of the longitudinal control pedal are both 0, the auxiliary brake module is recovered, the automobile is recovered to be in the single-pedal driving mode, and then the driver can normally drive the automobile, steps on the longitudinal control pedal to accelerate the automobile and releases the longitudinal control pedal to decelerate the automobile.
The preferred embodiments of the present invention have been described in detail with reference to the accompanying drawings, however, the present invention is not limited to the specific details of the above embodiments, and various simple modifications can be made to the technical solution of the present invention within the technical idea of the present invention, and these simple modifications are within the protective scope of the present invention.
It should be noted that the various technical features described in the above embodiments can be combined in any suitable manner without contradiction, and the invention is not described in any way for the possible combinations in order to avoid unnecessary repetition.
Claims (6)
1. A machine learning-based single-pedal wire-controlled chassis automobile auxiliary braking method is characterized by comprising the following steps:
s1: sample data acquisition: selecting 100 male testers and 100 female testers with the ages of C2 and above and the ages of the male testers and the female testers which are evenly distributed between 18 and 60 years old to carry out a real vehicle test in a driving field, wherein the real vehicle adopts an acceleration and braking integrated longitudinal control pedal to replace a traditional acceleration pedal, a driver steps on the longitudinal control pedal to accelerate, the longitudinal control pedal is released to decelerate through regenerative braking, and a brake pedal consistent with the traditional vehicle is arranged on the left side of the longitudinal control pedal and used for emergency braking; the method comprises the steps of collecting information of age, driving age, height and weight of each tester before a test is started, and then testing each tester under three working conditions of stable acceleration, rapid acceleration and longitudinal control pedal misoperation when encountering an obstacle, wherein 10 times of testing are arranged in each working condition, and the three working conditions randomly appear in the test process; monitoring the speed and the distance between an automobile and a front obstacle in real time in the test process, and acquiring the maximum value of the angular acceleration of a longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of a steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel of all testers under the working conditions that the testers accelerate suddenly and the longitudinal control pedal is operated by the obstacles in a wrong way;
s2: constructing a driver behavior model: establishing an end-to-end prediction model of the age, the driving age, the height, the weight, the speed of the driver and the distance between the automobile and a front obstacle, which is related to the maximum value of the angular acceleration of a longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of a steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the rotation angle of the steering wheel, the maximum value of the rotation angular speed of the steering wheel and the maximum value of the rotation angular acceleration of the steering wheel, based on the test data, wherein the end-to-end prediction model is established by a support vector regression model;
taking the age, driving age, height, weight, speed and distance between an automobile and a front obstacle in sample data as input variables of a support vector regression model, and taking the maximum value of the angular acceleration of a longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of a steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular speed of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel in the sample data as output variables of the support vector regression model; sending the sample data after normalization processing into a support vector regression model for training, predicting the maximum value of angular acceleration of a longitudinal control pedal, the maximum value of pressure of the longitudinal control pedal, the maximum value of hand holding force of a steering wheel, the maximum value of torque of the steering wheel, the maximum value of corner of the steering wheel, the maximum value of rotating angular speed of the steering wheel and the maximum value of rotating angular acceleration of the steering wheel by using the learned high-dimensional mapping relation, and providing a data basis for establishing a judging logic that a driver mistakenly steps on the pedal so as to mistakenly accelerate;
s3: monitoring whether the driver operates the longitudinal control pedal in a misoperation mode in real time: collecting the age, the driving age, the height and the weight of a driver before the driver starts to drive the automobile, monitoring the speed of the automobile and the distance between the automobile and a front obstacle in real time in the driving process of the automobile, and monitoring the angular acceleration a of a longitudinal control pedal, the pressure h of the longitudinal control pedal, the hand holding force f of a steering wheel, the torque m of the steering wheel, the turning angle theta of the steering wheel, the rotational angular speed omega of the steering wheel and the rotational angular acceleration beta of the steering wheel; meanwhile, the maximum value a of the angular acceleration of the longitudinal control pedal of the driver under the condition of rapid acceleration is calculated through a support vector regression model1Longitudinal control of maximum pedal pressure h1Maximum hand holding force f of steering wheel1Steering wheel torque maximum m1Maximum steering wheel angle θ1Maximum value of angular velocity ω of steering wheel rotation1And maximum value of angular acceleration of steering wheel1And maximum angular acceleration a of the longitudinal control pedal under the condition that the longitudinal control pedal is operated by mistake in case of an obstacle2Longitudinal control of maximum pedal pressure h2Maximum hand holding force f of steering wheel2Steering wheel torque maximum m2Maximum steering wheel angle θ2Maximum value of angular velocity ω of steering wheel rotation2And maximum value of angular acceleration of steering wheel2；
When any one of the following two conditions occurs, the system judges that the driver operates the longitudinal control pedal by mistake and starts the auxiliary brake module:
the first condition is as follows: the following seven conditions simultaneously satisfy four and more:
①a1≤a＜a2；
②h1≤h＜h2；
③f1≤f＜f2；
④m1≤m＜m2；
⑤θ1≤θ＜θ2；
⑥ω1≤ω＜ω2；
⑦β1≤β＜β2；
case two: any one of the following seven conditions is satisfied:
①a≥a2；
②h≥h2；
③f≥f2；
④m≥m2；
⑤θ≥θ2；
⑥ω≥ω2；
⑦β≥β2；
the auxiliary brake module starts to work after the system judges that a driver mistakenly operates the longitudinal control pedal, the automobile is changed from a single-pedal driving mode to a braking mode, when the driver downwards steps on the longitudinal control pedal, the automobile receives a braking signal, a braking execution mechanism starts to execute, an acceleration signal is cut off at the same time, when the pressure and the speed of the longitudinal control pedal are both 0, the auxiliary brake module is recovered, the automobile is recovered to the single-pedal driving mode, and then the driver can normally drive the automobile, steps on the longitudinal control pedal to accelerate the automobile and releases the deceleration of the longitudinal control pedal.
2. The machine learning-based single-pedal chassis-by-wire automobile auxiliary braking method according to claim 1, wherein in the step S2, the method comprises the following steps:
integrating the sample data into a data set D { (x)1,y1),(x2,y2)...,(xn,yn)}，
The training samples are mapped from a low-dimensional space to a high-dimensional space through nonlinear mapping, and a linear regression model established in the high-dimensional space can be expressed as the following equation:
f(x)＝w·Φ(x)+b
where x is an input variable, Φ (x) is a non-linear function mapping x to a high-dimensional linear space, w is a weight vector, b is an offset,
in order to minimize the regression error, the objective function of the vector regression algorithm is supported, and the following equation can be expressed:
wherein, CpIs a penalty coefficient which represents the penalty degree of the model to the sample with the error larger than epsilon in the training process, lεIs an epsilon-insensitive loss function, epsilon represents an insensitive loss coefficient, the smaller epsilon represents the smaller error requirement of the regression function, and lεThe expression can be expressed as the following equation:
wherein z represents the error between the fitting value and the true value of the support vector regression algorithm;
when the data does not conform to lε(z) constraint, introducing a relaxation variable deltai，δi *To correct for the irregular factor, the following equation can then be derived:
by introducing lagrange multiplier alphai、αi *Simplifying the calculation, converting the above formula into the solution of alphai，αi *The dual problem of (2):
wherein, K (x)i,xj) The method selects an RBF kernel function, and the RBF kernel function is defined as the following equation:
K(xi,xj)＝exp(-γ||xi-xj||2)
wherein gamma is a nuclear parameter;
according to the carlo-kun-tower condition, the regression function f (x) obtained by solving can be expressed as:
based on the above method, the driver behavior model can be abstracted as:
y＝f(x|(Cp,ε,γ))。
3. according to claimThe machine-learning-based single-pedal drive-by-wire chassis automobile auxiliary braking method of claim 2, characterized in that the adaptive particle swarm optimization is used to support three hyper-parameters, namely penalty coefficient C, of the vector regression modelpAnd optimizing a kernel parameter gamma and an insensitive loss coefficient epsilon.
4. The machine learning-based single-pedal drive-by-wire chassis automobile auxiliary braking method according to claim 3, characterized in that the mean absolute percentage error MAPE capable of directly reflecting regression performance is selected as the fitness function fitness of the adaptive particle swarm optimization, namely:
where n is the number of sample data, yiIs the predicted value, f (x)i) Are experimental values.
5. The machine-learning-based single-pedal chassis-by-wire automobile auxiliary braking method according to claim 1, wherein sample data is normalized in step S2 according to the following equation:
wherein, XminIs the minimum value of the sample data; xmaxIs the maximum value of the sample data; x is sample data; x' is normalized data in the range of 0,1]；
After the sample data is normalized, dividing the normalized data into two parts, wherein 80% of the normalized data is classified as a training data set, and 20% of the normalized data is classified as a testing data set.
6. The machine learning-based single-pedal chassis-by-wire automotive auxiliary braking method according to claim 1, characterized in that training is used in step S2The obtained support vector regression model predicts the maximum value of the angular acceleration of the longitudinal control pedal, the maximum value of the pressure of the longitudinal control pedal, the maximum value of the hand holding force of the steering wheel, the maximum value of the torque of the steering wheel, the maximum value of the turning angle of the steering wheel, the maximum value of the turning angular velocity of the steering wheel and the maximum value of the turning angular acceleration of the steering wheel, and adopts mean square error MSE and a decision coefficient R2And evaluating the prediction result of the model:
wherein,
is the average of the predicted values and is,
is the average of the experimental values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210179569.3A CN114407847B (en) | 2022-02-25 | 2022-02-25 | Single-pedal drive-by-wire chassis automobile auxiliary braking method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210179569.3A CN114407847B (en) | 2022-02-25 | 2022-02-25 | Single-pedal drive-by-wire chassis automobile auxiliary braking method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114407847A true CN114407847A (en) | 2022-04-29 |
| CN114407847B CN114407847B (en) | 2023-11-17 |
Family
ID=81261177
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210179569.3A Active CN114407847B (en) | 2022-02-25 | 2022-02-25 | Single-pedal drive-by-wire chassis automobile auxiliary braking method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114407847B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102010045694A1 (en) * | 2010-09-16 | 2012-03-22 | Daimler Ag | Method for preventing collision of e.g. motor car with obstacle during parking in tight parking lot, involves determining whether collision of vehicle is threatened in predeterminable period based on application of simulation evaluation |
| CN106274483A (en) * | 2016-11-10 | 2017-01-04 | 合肥工业大学 | The Vehicular automatic driving switching device differentiated based on driving behavior of diverting one's attention and method |
| KR20170137427A (en) * | 2016-06-03 | 2017-12-13 | 현대자동차주식회사 | Safety apparatus and method for operating pedal of autonomous driving vehicle |
| CN108909526A (en) * | 2018-06-26 | 2018-11-30 | 北京新能源汽车股份有限公司 | Single-pedal driving mode deceleration control method and device and electric automobile |
| CN109334564A (en) * | 2018-09-11 | 2019-02-15 | 南京航空航天大学 | A kind of automobile active safety early warning system of anticollision |
| CN112477865A (en) * | 2020-12-02 | 2021-03-12 | 南京航空航天大学 | Regenerative braking control method based on driver braking behavior prediction |
| CN112622614A (en) * | 2020-12-18 | 2021-04-09 | 的卢技术有限公司 | Single-pedal control system and control method thereof |
| CN113386768A (en) * | 2021-08-02 | 2021-09-14 | 合肥工业大学 | Single-pedal nonlinear model prediction control method for pure electric vehicle |
| US20210362718A1 (en) * | 2018-05-22 | 2021-11-25 | Yanmar Power Technology Co., Ltd. | Work Vehicle |
-
2022
- 2022-02-25 CN CN202210179569.3A patent/CN114407847B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102010045694A1 (en) * | 2010-09-16 | 2012-03-22 | Daimler Ag | Method for preventing collision of e.g. motor car with obstacle during parking in tight parking lot, involves determining whether collision of vehicle is threatened in predeterminable period based on application of simulation evaluation |
| KR20170137427A (en) * | 2016-06-03 | 2017-12-13 | 현대자동차주식회사 | Safety apparatus and method for operating pedal of autonomous driving vehicle |
| CN106274483A (en) * | 2016-11-10 | 2017-01-04 | 合肥工业大学 | The Vehicular automatic driving switching device differentiated based on driving behavior of diverting one's attention and method |
| US20210362718A1 (en) * | 2018-05-22 | 2021-11-25 | Yanmar Power Technology Co., Ltd. | Work Vehicle |
| CN108909526A (en) * | 2018-06-26 | 2018-11-30 | 北京新能源汽车股份有限公司 | Single-pedal driving mode deceleration control method and device and electric automobile |
| CN109334564A (en) * | 2018-09-11 | 2019-02-15 | 南京航空航天大学 | A kind of automobile active safety early warning system of anticollision |
| CN112477865A (en) * | 2020-12-02 | 2021-03-12 | 南京航空航天大学 | Regenerative braking control method based on driver braking behavior prediction |
| CN112622614A (en) * | 2020-12-18 | 2021-04-09 | 的卢技术有限公司 | Single-pedal control system and control method thereof |
| CN113386768A (en) * | 2021-08-02 | 2021-09-14 | 合肥工业大学 | Single-pedal nonlinear model prediction control method for pure electric vehicle |
Non-Patent Citations (2)
| Title |
|---|
| 宗长富;王畅;何磊;郑宏宇;张泽星;: "基于双层隐式马尔科夫模型的驾驶意图辨识", 汽车工程, no. 08 * |
| 段立飞;高振海;王德平: "驾驶员对汽车方向的自适应控制行为建模", 机械工程学报, vol. 47, no. 8 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114407847B (en) | 2023-11-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11731611B2 (en) | Vehicle stability control method and device | |
| CN108297877B (en) | Control method for vehicle, system and device | |
| CN109910617B (en) | Method for diagnosing failure fault of distributed hub motor driven vehicle | |
| EP2177413B1 (en) | Vehicle control system | |
| CN106364365A (en) | Method and device for processing double-path accelerator pedal signals | |
| CN109606460B (en) | Steering wheel connection decision method of double-rudder driving automobile based on steer-by-wire | |
| US10807634B2 (en) | Control method for electric power steering and control system for electric power steering | |
| CN107585207B (en) | A kind of vehicle line traffic control four-wheel steering system and its control method | |
| CN113547922B (en) | Pure electric vehicle energy recovery torque exit slope control method | |
| CN106864436A (en) | A kind of automatic driving vehicle brakes and braking method | |
| CN114802205A (en) | System and method for vehicle turning radius reduction | |
| CN114407847A (en) | Single-pedal wire-controlled chassis automobile auxiliary braking method based on machine learning | |
| CN103253274B (en) | First driver expects the method and control system of the credibility Analysis of sensor | |
| CN112918443A (en) | Brake control method, device, equipment and storage medium | |
| CN116605052A (en) | Torque control method, device and equipment for new energy automobile and storage medium | |
| US11724739B2 (en) | Vehicle actuation commands to affect transient handling | |
| CN116610022A (en) | Universal actuator with customized local rollback | |
| US11625959B2 (en) | System and method for determining backlash in driving system of vehicle | |
| CN114670668B (en) | Torque control method and device and electric automobile | |
| CN114889604A (en) | Regenerative braking control method of hybrid power vehicle based on electronic hydraulic braking system | |
| Boyraz et al. | Active accident avoidance case study: integrating drowsiness monitoring system with lateral control and speed regulation in passenger vehicles | |
| CN112733887A (en) | Method for detecting fault of hub motor of electric vehicle driven by supervision data | |
| Zhu et al. | Research on energy regenerative braking of electric vehicle based on functional safety analysis | |
| CN111605411A (en) | Active damping anti-shake control method and system of cross-controller and electric automobile | |
| US12001189B2 (en) | Systems and methods for improved manufacturing diagnostics |
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 |