CN113505545B - Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion - Google Patents
Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion Download PDFInfo
- Publication number
- CN113505545B CN113505545B CN202110686551.8A CN202110686551A CN113505545B CN 113505545 B CN113505545 B CN 113505545B CN 202110686551 A CN202110686551 A CN 202110686551A CN 113505545 B CN113505545 B CN 113505545B
- Authority
- CN
- China
- Prior art keywords
- pneumatic
- error
- point
- objective optimization
- pantograph
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Hardware Design (AREA)
- Computational Mathematics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Automation & Control Theory (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Fluid Mechanics (AREA)
- Mathematical Physics (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a track traffic vehicle pneumatic multi-objective optimization method based on an improved point adding criterion, in the aspect of proxy model construction, the improved comprehensive local/global search point adding criterion can add 1 improved expected adding point in each generation, and can also add a plurality of verification points in a Pareto solution set obtained by multi-objective optimization calculation to a next generation proxy model, and the proxy model with higher precision can be established by carrying out less iterations. And finally, the multi-objective optimization calculation result of the previous generation agent model is added to the next generation, so that the establishment of the agent model is fused with the multi-objective optimization, and the optimization calculation efficiency is improved.
Description
Technical Field
The invention relates to the field of aerodynamics, in particular to a track traffic vehicle pneumatic multi-objective optimization method based on an improved point adding criterion.
Background
The running speed of rail transit vehicles is increasing day by day, and taking high-speed trains (including high-speed maglev trains) as an example, the running speed of the rail transit vehicles is increased from 200km/h to 350 km/h-400 km/h, and the increase of the speed also aggravates the problems in pneumatic performance: the ratio of the pneumatic resistance to the total resistance of the train operation is increased sharply; the increase of the lift force of the tail car affects the traction efficiency of the train, and derailment accidents can be caused under special environment; for a magnetic suspension train, the overlarge aerodynamic lift force easily causes the failure of a suspension system, so that bottom equipment collides with a track beam, and the running safety of the train is influenced; aerodynamic noise also increases dramatically with increasing speed. Therefore, the aerodynamic performance optimization needs to be carried out on the appearance of the high-speed train, wherein the aerodynamic performance of the train is greatly influenced by the streamline position of the head of the train, namely the shape of the head of the train. When the optimization design of the head shape of the high-speed train is carried out, the aerodynamic resistance, the aerodynamic noise and the aerodynamic lift of the tail train need to be comprehensively considered, but the goals may conflict with each other, so the optimization design of the head shape of the train is a multi-objective appearance optimization problem. The traditional optimization design of the train head shape generally includes that a plurality of head shape schemes are firstly drawn up, and then optimization is carried out through wind tunnel test and numerical calculation. And then, a train head type multi-objective optimization method based on the proxy model is developed, but the traditional proxy model method has the defects of low model building efficiency and more required sample points. Therefore, the high-speed train head shape is optimized by adopting an improved comprehensive/local search plus-point criterion pneumatic performance agent model in the rail transit field and multi-objective optimization, and the high-speed train head shape with better comprehensive pneumatic performance is obtained.
The pantograph transmits the electric energy of a contact net to a high-speed train through the sliding plate, and proper pantograph-catenary contact force is the basis for safely transmitting the electric energy. The friction force can be increased when the contact force of the bow net is too large, and the abrasion of the sliding plate and a contact net line is aggravated; if the contact pressure is too small, the contact resistance of the pantograph-catenary can be increased, and pantograph-catenary electric arcs can be generated; when the contact force is zero, the pantograph is off-line, the pantograph-catenary contact is broken, arc discharge occurs, the contact line and the sliding plate are ablated, electric elements are damaged, and the driving safety of the train is seriously affected. The lifting force of the pantograph consists of a static lifting force and a pneumatic lifting force of the pantograph per se, and is an important component of the pantograph-catenary contact force (in addition, one part of the static lifting force is a dynamic component force caused by vibration), the static lifting force is provided by an air bag and can be adjusted by a control motor, and the pneumatic lifting force of the pantograph is related to the structure of the pantograph per se, the running speed of a train and the like and cannot be directly controlled manually. The pneumatic lifting force of the pantograph is the force which is vertically acted on a contact net and generated by a pantograph head under the comprehensive action of airflow on the aerodynamic force of each part of the pantograph, the pantograph has two running states of open running and closed running, the pneumatic lifting force of the pantograph in the two running states is inconsistent due to the longitudinal asymmetry of the pantograph structure, and an air bag system is required to be adjusted to adjust the static lifting force when a train turns the running direction. Therefore, the method has the advantages that the pneumatic lifting force of the pantograph is optimized, the pneumatic lifting force of the operation working condition of the opening and closing port is adjusted to be close, the dependence on a static lifting force active control system can be reduced, the fatigue damage of an air bag system is reduced, the occurrence rate of pantograph net accidents is reduced, and the operation safety of a high-speed train is guaranteed. In addition, under different running speeds, the pneumatic lifting force of the pantograph should meet the standard under the corresponding speed grade, so the pneumatic lifting force of the pantograph is optimized by adopting an improved traffic field pneumatic performance proxy model with comprehensive/local search and point adding criteria and a multi-objective optimization method, the pneumatic lifting force of the pantograph meets the standard requirement, and the difference of the pneumatic lifting force of the opening and closing port is small.
In the aspect of optimizing the head shape of the high-speed train, the traditional optimization design of the head shape of the train is generally to draw up a plurality of head shape schemes, and then optimization is carried out through wind tunnel tests and numerical calculation, so that the cost is high, and the optimization period is long. The train head type multi-objective optimization method based on the agent model is low in model building efficiency and needs more sample points.
In the aspect of the pneumatic lifting force of pantograph, under the inconsistent condition of the pneumatic lifting force of pantograph opening and closing mouth operation condition, the need behind the high-speed train turning direction of operation adjusts the pantograph gasbag system and lifts lifting force in order to change static state, and this can increase train operation cost, and aggravates the fatigue damage of gasbag system, causes the bow net accident even, influences the operation security of train.
In the aspect of proxy model construction, the traditional point adding criterion can only calculate 1 adding point each time, a next-generation proxy model is constructed after a sample set is added, and if a proxy model with higher precision is to be constructed, multiple point adding is needed to update a multi-generation proxy model. Finally, the traditional multi-objective optimization calculation is to perform multi-objective calculation after a proxy model is established, if the precision is not enough, points are continuously added to establish a next generation proxy model, and then the multi-objective optimization calculation is performed, wherein the two are independent, so that the optimization efficiency is low.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a track traffic vehicle pneumatic multi-objective optimization method based on an improved point adding criterion.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
a pneumatic multi-objective optimization method of a rail transit vehicle based on an improved point adding criterion comprises the following steps:
s1, taking aerodynamic parameters of the rail vehicle as independent variables, and sampling the independent variables by utilizing a Latin hypercube sampling method to obtain initial sample points;
s2, establishing a primary proxy model of the rail transit aerodynamic performance according to the initial sample points obtained in the step S1, and calculating the required improvement expectation addition points by using an improvement expectation method;
s3, merging the expected improving adding points obtained in the step S2 and the initial sample points obtained in the step S1 into a sample set, and establishing a first generation proxy model according to the sample set;
s4, performing multi-objective optimization calculation on the first generation agent model established in the step S3 by adopting an NSGA-II algorithm to obtain a leading edge solution set, selecting a plurality of points from the solution set as verification points, and calculating a true value of the verification points;
s5, calculating a predicted value by using the first generation agent model established in the step S3, comparing the predicted value with the true value obtained in the step S4, judging whether the error between the predicted value and the true value meets the requirement of a set threshold, and if so, finishing improving the expected addition point; if not, the method returns to step S2 to re-add the improvement expected addition point, and iterates the proxy model until the error meets the set threshold requirement.
Further, in step S1, the independent variables include a pantograph pneumatic lifting force parameter of the rail train and a head control line parameter of the high-speed train.
Further, when the independent variable is a pneumatic lifting force parameter of a pantograph of the rail train, the optimization targets are that the pneumatic lifting force difference of an opening and closing port of the pantograph is minimum, the pneumatic resistance of the whole pantograph of the opening and closing port is minimum, and the pneumatic lifting force meets the standard constraint.
Further, when the independent variable is a rail train pantograph pneumatic lifting force parameter, the error requirements between the truth value of the verification point and the predicted value of the proxy model are that the error of the opening pneumatic lifting force is less than 3N, the error of the opening and closing opening pneumatic lifting force difference is less than 5N, and the pneumatic resistance and error of the whole pantograph of the opening and closing opening are less than 1% simultaneously.
Further, when the independent variable is a high-speed train head control line parameter, the optimization target is that the whole high-speed train aerodynamic resistance, aerodynamic noise and tail train aerodynamic lift are minimum.
Further, when the independent variable is a parameter of a control line of a head of the high-speed train, the error requirement between the true value of the verification point and the predicted value of the proxy model meets the requirements that the aerodynamic resistance error of the whole train is less than 1%, the aerodynamic noise error is less than 3% and the aerodynamic lift error of the tail train is less than 3%.
The invention has the following beneficial effects:
1. compared with the traditional point adding criterion, the point adding criterion can additionally add a plurality of adding points, and the proxy model with higher precision can be established by carrying out fewer iterations.
2. And adding the multi-objective optimization calculation result of the previous generation agent model to the next generation, so that the establishment of the agent model and the multi-objective optimization are fused, and the multi-objective optimization calculation efficiency can be improved.
3. The pantograph pneumatic lifting force under the operation working condition of the opening and closing port can be optimized to meet the standard requirement and have small difference.
4. The head type scheme of the high-speed train with better comprehensive pneumatic performance can be obtained.
Drawings
FIG. 1 is a process of proxy model construction and multi-objective optimization in accordance with the present invention.
FIG. 2 is Polonif1、f2The two dotting criteria of the function converge the history.
FIG. 3 is a Pareto front obtained by multi-objective optimization calculation of a Poloni function proxy model established by improving a point adding criterion, and is compared with a real front.
FIG. 4 shows CT-E1 g1、g2The two dotting criteria of the function converge the history.
FIG. 5 is a Pareto front edge obtained by multi-objective optimization calculation of a CT-E1 function proxy model established by improving a point adding criterion compared with a real front edge.
FIG. 6 is a Pareto front edge solution set obtained by multi-objective optimization calculation of the pantograph pneumatic lifting force.
Fig. 7 is an error convergence course in the iterative update process of the pantograph pneumatic lifting force proxy model.
FIG. 8 shows the variables and their variation intervals in the multi-objective optimization design of the high-speed train head.
FIG. 9 is a Pareto front solution set obtained by high-speed train head type multi-objective optimization calculation.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
A rail transit vehicle pneumatic multi-objective optimization method based on an improved point adding criterion is shown in figure 1 and comprises the following steps:
s1, taking aerodynamic parameters of the rail vehicle as independent variables, and sampling the independent variables by utilizing a Latin hypercube sampling method to obtain initial sample points;
in the embodiment, the independent variables comprise a pantograph pneumatic lifting force parameter of the rail train and a locomotive control line parameter of the high-speed train,
when the parameters of the pneumatic lifting force of the pantograph are used as multi-objective optimization, the diameter of an upper arm rod, the diameter of a lower arm rod and the angle of an air deflector of the pantograph are used as independent variables, the pneumatic lifting force difference of an opening and closing port of the pantograph is minimum, the pneumatic resistance of the whole pantograph of the opening and closing port is minimum, and the pneumatic lifting force meets the standard and is constrained to perform multi-objective optimization;
when the head shape of the high-speed train is taken as multi-objective optimization, 5 design variables v are selected1、v2、v3、v4And v 55 control lines corresponding to the train head shape, 5 control lines being nose tip height control lines C1And a height control line C of a cab window2Horizontal outer contour control line C3Middle auxiliary control line C4And a middle auxiliary control line C5，v1、v2And v3Represents the maximum distance of curve deformation, v4And v5The deformation ratio of the original curve is shown, and the variation intervals of the design variables are shown in fig. 8. And carrying out multi-target optimization on the pneumatic performance of the high-speed train by taking the minimum aerodynamic resistance, aerodynamic noise and tail vehicle aerodynamic lift of the whole high-speed train as an optimization target.
S2, establishing a primary proxy model of the rail transit aerodynamic performance according to the initial sample points obtained in the step S1, and calculating the required improvement expectation addition points by using an improvement expectation method;
the proxy model is simply described as a mapping relation y ═ f (x) for converting an input vector x into a response y, which can be understood as a black box problem, and the establishment of the proxy model is to find an optimal approximation function for the black box mapping y ═ f (x) based on the known input vector x and the response y
Approximation function
The establishment of the model can utilize polynomial fitting, namely a polynomial proxy model, or a response surface model. Approximation function can also be used
Built as a combination of a plurality of simple basis functions, i.e. pathsA basis function (RBF) method. The proxy model is built based on Gauss basis functions in radial basis functions, and as a special Gauss basis function, a variable theta is introduced into the basis functions of the Kriging proxy model, the bandwidth of the basis functions can be changed along with the change of the variable, the precision is improved compared with the proxy model built by the general Gauss basis functions, and the basis functions of the Kriging proxy model are as follows:
in the formula: k is the number of design variables and p is fixed to 2 in this embodiment.
For design variable x ═ x(1),x(2),…,x(n)}TAnd it
True response value y ═ y(1),y(2),…,y(n)}TIn the establishment process of the Kriging agent model, it is assumed that the response comes from a random process:
Y＝(Y(x(1)),Y(x(2)),…,Y(x(n))T
the mean value of the random field is 1 mu, and the random variables are related by using a basis function:
and then establishing a correlation matrix of all observation points by using the following formula:
then for an unknown vector x', its response value can be expressed as:
in the formula: psi is the correlation vector of the sample point x and the prediction point x
A large number of sample points are needed for establishing a high-precision proxy model, so that the proxy model based on the point adding strategy obtains the sample points needing to be added through some point adding criteria besides the initial sample points obtained by sampling.
If an accurate optimal solution is obtained in the optimization process, the prediction precision of the agent model near the optimal solution can be only improved; if the global precision of the proxy model is to be ensured, a point adding strategy for improving the global precision is required to be adopted.
The surrogate model of the gaussian process can provide an error estimate for the predicted value, such as the Kriging model, from which the location where the model prediction inaccuracy is greatest can be determined, and then adding sample points at that location, a feature that is also a significant advantage of the Kriging model.
The Mean Square Error (MSE) of the predicted value of the Kriging model can be expressed as:
in the formula: psi is the n-th order correlation matrix for all sample points,
σ is the standard deviation of the correlation vector between the sample point and the predicted point. The third term in parentheses represents an uncertainty estimate for the mean μ, which is very small and generally ignored.
When searching for sample points needing to be added, the optimal solution y can be improvedminThe point with the maximum probability is used as an adding point, and
regarded as a random process, it can be calculated that y is greater thanminImprovement of (I) yminProbability of existence, i.e. improvement of probability
Upon determining the mean value as
And variance of
On the premise that the expected improvement value can be calculated
In the formula: phi and phi are respectively a standard normal cumulative distribution function and a standard normal probability density function.
If an error function is used, it can be written as:
s3, merging the expected improving adding points obtained in the step S2 and the initial sample points obtained in the step S1 into a sample set, and establishing a first generation proxy model according to the sample set;
s4, performing multi-objective optimization calculation on the first generation agent model established in the step S3 by adopting an NSGA-II algorithm to obtain a leading edge solution set, selecting a plurality of points from the solution set as verification points, and calculating a true value of the verification points;
s5, calculating a predicted value by using the first generation agent model established in the step S3, comparing the predicted value with the true value obtained in the step S4, judging whether the error between the predicted value and the true value meets the requirement of a set threshold, and if so, finishing improving the expected addition point; if not, the method returns to step S2 to re-add the improvement expected addition point, and iterates the proxy model until the error meets the set threshold requirement.
And establishing an agent model for the constraint function by adopting the sample points when the target function agent model is established, calculating the probability that the predicted value is greater than the constraint value by using the constraint function agent model, and calculating the probability that the design point meets the constraint similarly to the calculation of the improved probability.
In the formula: g is a constraint function, gminFor constraint boundaries, F is a feasible degree, i.e., F (x) g (x) -gminG (x) is g (x) the corresponding random variable,
is the variance of the proxy model.
The random processes of the target and the constraint function are mutually independent, so that the improvement probability of the target function and the probability of meeting the constraint can be combined to be used as an adding point to improve the current optimal solution and meet the probability of the constraint function
P[I(x)∩F(x)]＝P[I(x)]P[F(x)]
E[I(x)]Multiplication P [ F (x)>gminConstraint improvement expectation dotting criterion can be derived
E[I(x)∩F(x)]＝E[I(x)]P[F(x)]。
Example 1
Take the multi-objective Poloni test function as an example:
Polonif1、f2the expression of the function is:
f1(x,y)＝-[1+(A1-B1)2+(A2-B2)2]
f2(x,y)＝-[(x+3)2+(y+1)2]
in the formula: x, y is larger than the maximum value (-pi, pi),
A1＝0.5sin1-2cos1+sin2-1.5cos2
A2＝1.5sin1-cos1+2sin2-0.5cos2
B1＝0.5sin x-2cos x+sin y-1.5cos y
B2＝1.5sin x-cos x+2sin y-0.5cos y
1) calculating the required improved expected addition point P by using the improved expected method for the initial sample point obtained by the Latin hypercube sampling method1。
2) Adding the improvement expectation point P1Combining the initial sample points with the initial sample points to form a sample set, establishing a generation 1 proxy model, performing multi-objective optimization calculation on the generation 1 model by adopting an NSGA-II algorithm to obtain a pareto front solution set, selecting 2 points from the solution set as verification points, and calculating the true values of the verification points.
3) And comparing the true values of the verification points with the predicted values of the agent model, wherein the error requirement is that the average difference between the predicted values and the true values of all the verification points is less than 0.4, and if the error between the prediction result of the agent model and the true values meets the requirement, the agent model and the multi-objective optimization calculation are considered to have higher accuracy, point adding is finished, and the optimization calculation is finished.
4) If the error does not meet the requirement, adding the 2 verification points into the sample set as optimization addition points, and calculating by using an improvement expectation method to obtain a required improvement expectation addition point P2Adding the improvement to the desired addition point P2And adding a sample set to establish a 2 nd generation proxy model, and returning to the step 3).
5) Establishing a proxy model according to the steps and performing multi-objective optimization, Polonif1、f2The function satisfies the error criterion after iteration to 3 rd generation, whereas using the conventional dotting criterion requires iteration to 10 th generation before the error criterion is satisfied. Polonif1、f2The convergence process of the two kinds of point adding criteria of the function is shown in fig. 2, and the Pareto leading edge and real leading edge pair obtained by establishing a proxy model and performing multi-objective optimization calculation by using the improved point adding criteria is shown in fig. 3.
Example 2
Taking the constrained multi-target CT-E1 test function as an example:
CT-E1 g1、g2the expression of the function is:
g1＝x1，
in the formula: x is the number of1∈(0.1，1)，x2E (0,5), the constraint relation is as follows:
9x1+x2≥6，9x1-x2≥1
1) the initial sample point obtained by the Latin hypercube sampling method is calculated by the improvement expectation method to obtain the required improvement expectation addition point T1。
2) Adding the improvement expectation to the point T1Combining the initial sample points with the initial sample points to form a sample set, establishing a generation 1 proxy model, performing multi-objective optimization calculation on the generation 1 model by adopting an NSGA-II algorithm to obtain a pareto front solution set, selecting 2 points from the solution set as verification points, and calculating the true values of the verification points.
3) And comparing the real values of the verification points with the predicted values of the agent model, wherein the error requirement is that the average relative error of the predicted values and the real values of all the verification points is less than 1%, and if the error of the prediction result and the real values of the agent model meets the requirement, the agent model and the multi-objective optimization calculation are considered to have higher accuracy, point adding is finished, and the optimization calculation is finished.
4) If the error does not meet the requirement, adding the 2 verification points serving as optimization addition points into the sample set, and calculating by using an improvement expectation method to obtain a required improvement expectation addition point T2Adding the improvement to the desired addition point T2And adding a sample set to establish a 2 nd generation proxy model, and returning to the step 3).
5) Establishing a proxy model according to the steps and performing multi-objective optimization, CT-E1 g1、g2The function satisfies the error criterion after iteration to 3 rd generation, while using the conventional dotting criterion requires iteration to 8 th generation to satisfy the error criterion. CT-E1 g1、g2The convergence process of the two kinds of point adding criteria of the function is shown in fig. 4, and the Pareto leading edge and real leading edge pair obtained by establishing a proxy model and performing multi-objective optimization calculation by using the improved point adding criteria is shown in fig. 5.
Example 3
Establishing a multi-objective optimization proxy model of the pneumatic lifting force of the pantograph, taking the diameter of an upper arm rod, the diameter of a lower arm rod and the angle of an air deflector of the pantograph as independent variables, taking the minimum pneumatic lifting force difference of an opening and closing port of the pantograph and the minimum pneumatic resistance of the whole pantograph of the opening and closing port as optimization targets, and taking the pneumatic lifting force meeting the standard as constraint to carry out multi-objective optimization.
1) Calculating the required improved expected addition point E by using the improved expected method for the initial sample point obtained by the Latin hypercube sampling method1。
2) Adding point E to the improvement expectation1Combining the initial sample points with the initial sample points to form a sample set, establishing a generation 1 proxy model, performing multi-objective optimization calculation on the generation 1 model by adopting an NSGA-II algorithm to obtain a pareto front solution set, selecting 2 points from the solution set as verification points, and calculating the true values of the verification points.
3) And comparing the true value of the verification point with the predicted value of the proxy model, wherein the error requirements are that the error of the opening pneumatic lifting force is less than 3N, the error of the opening and closing port pneumatic lifting force difference is less than 5N, and the pneumatic resistance and error of the whole arch of the opening and closing port are less than 1%. And if the error between the prediction result of the agent model and the true value meets the requirement, considering that the accuracy of the agent model and the multi-target optimization calculation is higher, ending the point adding and ending the optimization calculation.
4) If the error does not meet the requirement, adding the 2 verification points into the sample set as optimization addition points, and calculating by using an improvement expectation method to obtain a required improvement expectation addition point E2Adding point E to the improvement2And adding a sample set to establish a 2 nd generation proxy model, and returning to the step 3).
5) And establishing a proxy model according to the steps and performing multi-objective optimization, wherein after iteration to the 4 th generation, the error standard is met, in order to further improve the precision, the 4 th generation verification point is added into the sample set to construct a 5 th generation proxy model, the Pareto leading edge obtained by multi-objective optimization calculation is shown in figure 6, 3 verification points are selected from the leading edge, and the error convergence course in the iterative updating process of the proxy model is shown in figure 7. All verification points in the 5 th generation meet the error requirement, and finally, the optimal pantograph scheme is obtained: the pneumatic lifting force meets the standard, and the difference value of the pneumatic lifting force under the operation condition of the opening and the closing is only 3.8N.
Example 4
Establishing a high-speed train head type multi-objective optimization proxy model, and selecting 5 design variables v1、v2、v3、v4And v 55 control lines corresponding to the train head shape, 5 control lines being nose tip height control lines C1And a height control line C of a cab window2Horizontal outer contour control line C3Middle auxiliary control line C4And a middle auxiliary control line C5，v1、v2And v3Represents the maximum distance of curve deformation, v4And v5The deformation ratio of the original curve is shown, and the variation intervals of the design variables are shown in fig. 8. And carrying out multi-target optimization on the pneumatic performance of the high-speed train by taking the minimum aerodynamic resistance, aerodynamic noise and tail vehicle aerodynamic lift of the whole high-speed train as an optimization target.
1) Calculating the required improvement expectation addition point H by using an improvement expectation method through the initial sample point obtained by the Latin hypercube sampling method1。
2) Adding the improvement to the desired addition point H1Combining the initial sample points with the initial sample points to form a sample set, establishing a generation 1 proxy model, performing multi-objective optimization calculation on the generation 1 model by adopting an NSGA-II algorithm to obtain a pareto front solution set, selecting 2 points from the solution set as verification points, and calculating the true values of the verification points.
3) And comparing the true value of the verification point with the predicted value of the proxy model, wherein the error requirements are that the pneumatic resistance error of the whole vehicle is less than 1%, the pneumatic noise error is less than 3% and the pneumatic lift error of the tail vehicle is less than 3% simultaneously. And if the error between the prediction result of the agent model and the true value meets the requirement, considering that the accuracy of the agent model and the multi-target optimization calculation is higher, ending the point adding and ending the optimization calculation.
4) If the error does not meet the requirement, adding the 2 verification points into the sample set as optimization addition points, and calculating by using an improvement expectation method to obtain a required improvement expectation addition point H2Adding the improvement to the desired addition point H2And adding a sample set to establish a 2 nd generation proxy model, and returning to the step 3).
5) And (3) establishing a proxy model according to the steps, performing multi-objective optimization, and after iteration to the 7 th generation, meeting an error standard, performing multi-objective optimization calculation to obtain a Pareto front edge as shown in FIG. 9, selecting 3 verification points from the front edge, and meeting an error requirement, so as to finally obtain an optimal high-speed train head type scheme, wherein the original model is reduced by 3.58% compared with the whole train aerodynamic resistance, the tail train aerodynamic lift is reduced by 5.67%, and the aerodynamic noise is reduced by 0.37%.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The principle and the implementation mode of the invention are explained by applying specific embodiments in the invention, and the description of the embodiments is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (6)
1. The rail transit vehicle pneumatic multi-objective optimization method based on the improved point adding criterion is characterized by comprising the following steps of:
s1, taking aerodynamic parameters of the rail vehicle as independent variables, and sampling the independent variables by utilizing a Latin hypercube sampling method to obtain initial sample points;
s2, establishing a primary proxy model of the rail transit aerodynamic performance according to the initial sample points obtained in the step S1, and calculating the required improvement expectation addition points by using an improvement expectation method, wherein,
the specific calculation mode of the improvement expectation method is as follows:
in the case where the mean and variance of the initial samples are determined, the improvement expectation method is calculated by:
wherein, E [ I (x)]In order to improve the expectation value,
is the average of the initial sample points and,
is the variance of the initial sample point, i (x) yminIn order to improve the optimal solution, phi is a standard normal cumulative distribution function and phi is a standard normal probability density function;
when using the error function, the calculation of the improvement expectation method is:
the improvement expectation addition point is the point with the maximum improvement optimal solution probability and is represented as:
s3, merging the expected improving adding points obtained in the step S2 and the initial sample points obtained in the step S1 into a sample set, and establishing a first generation proxy model according to the sample set;
s4, performing multi-objective optimization calculation on the first generation agent model established in the step S3 by adopting an NSGA-II algorithm to obtain a leading edge solution set, selecting a plurality of points from the solution set as verification points, and calculating a true value of the verification points;
s5, calculating a predicted value by using the first generation agent model established in the step S3, comparing the predicted value with the true value obtained in the step S4, judging whether the error between the predicted value and the true value meets the requirement of a set threshold value, if so, ending the process, and outputting the optimal pneumatic force parameter; if not, the verification point selected in the step S3 is taken as an optimization adding point and put into the initial sample point, and the step S2 is returned until the error between the selected verification point and the predicted value meets the requirement of the set threshold.
2. The rail transit vehicle pneumatic multi-objective optimization method based on the improved dotting criterion of claim 1, wherein the independent variables in the step S1 comprise pantograph pneumatic lifting force parameters of rail trains or high-speed train head control line parameters.
3. The rail transit vehicle pneumatic multi-objective optimization method based on the improved dotting criterion is characterized in that when the independent variable is a rail train pantograph pneumatic lifting force parameter, the optimization objectives are that the pantograph opening and closing port pneumatic lifting force difference is minimum, the opening and closing port whole-pantograph pneumatic resistance is minimum, and the pneumatic lifting force meets the standard constraint.
4. The rail transit vehicle pneumatic multi-objective optimization method based on the improved dotting criterion is characterized in that when the independent variable is a rail train pantograph pneumatic lifting force parameter, the error requirement between the truth value of the verification point and the predicted value of the proxy model is that the error of the opening pneumatic lifting force is less than 3N, the error of the opening and closing pneumatic lifting force difference is less than 5N, and the pneumatic resistance and the error of the opening and closing whole pantograph are less than 1% at the same time.
5. The rail transit vehicle pneumatic multi-objective optimization method based on the improved dotting criterion is characterized in that when the independent variable is a high-speed train head control line parameter, the optimization objective is to minimize the whole train pneumatic resistance, the pneumatic noise and the tail train pneumatic lift of the high-speed train.
6. The rail transit vehicle pneumatic multi-objective optimization method based on the improved dotting criterion of claim 5, wherein when the independent variable is a high-speed train head control line parameter, the error requirement between the true value of the verification point and the predicted value of the proxy model simultaneously meets the requirements that the whole train pneumatic resistance error is less than 1%, the pneumatic noise error is less than 3%, and the tail train pneumatic lift error is less than 3%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110686551.8A CN113505545B (en) | 2021-06-21 | 2021-06-21 | Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110686551.8A CN113505545B (en) | 2021-06-21 | 2021-06-21 | Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113505545A CN113505545A (en) | 2021-10-15 |
| CN113505545B true CN113505545B (en) | 2022-04-15 |
Family
ID=78010549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110686551.8A Active CN113505545B (en) | 2021-06-21 | 2021-06-21 | Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113505545B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114880775B (en) * | 2022-05-10 | 2023-05-09 | 江苏大学 | Feasible domain searching method and device based on active learning Kriging model |
| CN114861458B (en) * | 2022-05-26 | 2023-07-14 | 西南交通大学 | Bridge traffic safety rapid evaluation method based on mapping relation proxy model |
| CN115034122B (en) * | 2022-08-12 | 2022-11-15 | 成都佰维存储科技有限公司 | Method and device for optimizing 3C product structure, readable storage medium and electronic equipment |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004304895A (en) * | 2003-03-31 | 2004-10-28 | Railway Technical Res Inst | Pantograph |
| WO2008112921A1 (en) * | 2007-03-14 | 2008-09-18 | Halliburton Energy Services, Inc. | Neural-network based surrogate model construction methods and applications thereof |
| CN102867098A (en) * | 2012-09-27 | 2013-01-09 | 西南交通大学 | Component-oriented high-speed train system dynamics computer modeling method |
| CN108664721A (en) * | 2018-05-03 | 2018-10-16 | 中南大学 | High-speed train head shape collaborative design method based on multi-subject design |
| CN111079228A (en) * | 2019-12-19 | 2020-04-28 | 中国人民解放军国防科技大学 | Pneumatic shape optimization method based on flow field prediction |
| CN112329140A (en) * | 2020-10-30 | 2021-02-05 | 北京理工大学 | Method for optimizing aerodynamics of variant aircraft based on improved position vector expectation improvement degree |
| CN112434448A (en) * | 2021-01-27 | 2021-03-02 | 中国人民解放军国防科技大学 | Proxy model constraint optimization method and device based on multipoint adding |
| CN112749495A (en) * | 2021-01-27 | 2021-05-04 | 中国人民解放军国防科技大学 | Multipoint-point-adding-based proxy model optimization method and device and computer equipment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110889202A (en) * | 2019-11-06 | 2020-03-17 | 长沙理工大学 | Gear transmission device multi-objective optimization design method based on self-adaptive agent model |
-
2021
- 2021-06-21 CN CN202110686551.8A patent/CN113505545B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004304895A (en) * | 2003-03-31 | 2004-10-28 | Railway Technical Res Inst | Pantograph |
| WO2008112921A1 (en) * | 2007-03-14 | 2008-09-18 | Halliburton Energy Services, Inc. | Neural-network based surrogate model construction methods and applications thereof |
| CN102867098A (en) * | 2012-09-27 | 2013-01-09 | 西南交通大学 | Component-oriented high-speed train system dynamics computer modeling method |
| CN108664721A (en) * | 2018-05-03 | 2018-10-16 | 中南大学 | High-speed train head shape collaborative design method based on multi-subject design |
| CN111079228A (en) * | 2019-12-19 | 2020-04-28 | 中国人民解放军国防科技大学 | Pneumatic shape optimization method based on flow field prediction |
| CN112329140A (en) * | 2020-10-30 | 2021-02-05 | 北京理工大学 | Method for optimizing aerodynamics of variant aircraft based on improved position vector expectation improvement degree |
| CN112434448A (en) * | 2021-01-27 | 2021-03-02 | 中国人民解放军国防科技大学 | Proxy model constraint optimization method and device based on multipoint adding |
| CN112749495A (en) * | 2021-01-27 | 2021-05-04 | 中国人民解放军国防科技大学 | Multipoint-point-adding-based proxy model optimization method and device and computer equipment |
Non-Patent Citations (2)
| Title |
|---|
| Research on aerodynamic shape optimization of trains with different dimensional design variables;Le Zhang ET AL.;《International Journal of Rail Transportation》;20200908;第1-23页 * |
| 基于近似模型的高速列车头型多目标优化设计;于梦阁，等;《机械工程学报》;20191231;第178-186页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113505545A (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113505545B (en) | Rail transit vehicle pneumatic multi-objective optimization method based on improved point adding criterion | |
| CN109764886A (en) | A kind of paths planning method | |
| WO2021223334A1 (en) | Method for iterative joint estimation of vehicle mass and road gradient on the basis of mmrls and sh-stf | |
| CN114442637B (en) | Unmanned vehicle local dynamic obstacle avoidance path planning method | |
| CN112389436B (en) | Safety automatic driving track changing planning method based on improved LSTM neural network | |
| CN104859650B (en) | A kind of vehicle yaw stability rolling optimization control method of Multiple Time Scales | |
| Yu et al. | MPC-based regional path tracking controller design for autonomous ground vehicles | |
| CN115328152B (en) | Unmanned vehicle local real-time obstacle avoidance path planning method based on improved artificial potential field method | |
| CN112068445B (en) | Integrated control method and system for path planning and path tracking of automatic driving vehicle | |
| CN112578672B (en) | Unmanned vehicle trajectory control system based on chassis nonlinearity and trajectory control method thereof | |
| CN115683145A (en) | Automatic driving safety obstacle avoidance method based on track prediction | |
| CN112668779A (en) | Preceding vehicle motion state prediction method based on self-adaptive Gaussian process | |
| Valero et al. | Assessment of the effect of energy consumption on trajectory improvement for a car-like robot | |
| CN107239628A (en) | A kind of uncertain locomotive simulation model system construction method based on dynamic time sequence figure | |
| CN111595354B (en) | EKF-SLAM algorithm of self-adaptive dynamic observation domain | |
| CN113064424A (en) | Unmanned vehicle path planning method for improving DDPG algorithm | |
| Yin et al. | Nonlinear model predictive control for path tracking using discrete previewed points | |
| Lin et al. | Adaptive prediction-based control for an ecological cruise control system on curved and hilly roads | |
| Domina et al. | Path following controller for autonomous vehicles | |
| CN115963855A (en) | Unpowered reentry aircraft landing area prediction method based on deep learning | |
| Wang et al. | A double-layered nonlinear model predictive control based control algorithm for local trajectory planning for automated trucks under uncertain road adhesion coefficient conditions | |
| Shi et al. | Local trajectory planning for autonomous trucks in collision avoidance maneuvers with rollover prevention | |
| CN115343950A (en) | Vehicle path tracking control method and control system suitable for complex road surface | |
| CN111857112B (en) | Automobile local path planning method and electronic equipment | |
| CN114228748B (en) | Geometric path generation-based humanized automatic driving track planning method |
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 |