CN115577615A - Automobile multi-body dynamics modeling method and system - Google Patents
Automobile multi-body dynamics modeling method and system Download PDFInfo
- Publication number
- CN115577615A CN115577615A CN202211116217.XA CN202211116217A CN115577615A CN 115577615 A CN115577615 A CN 115577615A CN 202211116217 A CN202211116217 A CN 202211116217A CN 115577615 A CN115577615 A CN 115577615A
- Authority
- CN
- China
- Prior art keywords
- model
- response
- response surface
- preset
- iteration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- 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
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Computational Mathematics (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Medical Informatics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a method and a system for modeling automobile multi-body dynamics, wherein the method comprises the steps of establishing an initial multi-body dynamics model according to a vibration model and a Lagrange equation of an automobile; establishing a precision index by a working condition transmission path analysis method; according to load spectrum data and iteration conditions acquired by each road condition, performing virtual iteration checking on the initial multi-body dynamic model, respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of precision indexes, and if so, taking the initial model as an optimal model; if any local model does not meet the requirement, continuing iteration until all local models meet the precision requirement, and then performing overall optimization; if the requirements are locally met and the initial multi-body kinetic model does not meet the requirements, carrying out overall optimization, designing an orthogonal Latin hypercube test and carrying out response surface optimization evaluation; and finally establishing an optimal multi-body dynamic model. The modeling accuracy, the reliability and the applicability are effectively improved.
Description
Technical Field
The invention relates to the field of multi-body dynamics modeling, in particular to a method and a system for modeling automobile multi-body dynamics.
Background
The automobile dynamics simulation is directly related to the operation stability, the whole smoothness and the reliability of the automobile. Meanwhile, the accuracy of the analysis result of the automobile ride comfort performance directly depends on the accuracy of the established dynamic simulation model. In order to obtain accurate dynamic response and improve the dynamic performance of the whole vehicle, the multi-body dynamic modeling of the whole vehicle is particularly important, the multi-body dynamic modeling is used for shortening the product design period, saving new product research and development funds and reducing development risks, and the modeling accuracy, real-time performance, applicability and credibility directly determine whether the model simulation can be applied to product research and development.
Although research and application of multi-body dynamics modeling have been developed for many years, due to continuous development and increasing complexity of the structure and function of a modeling object, the dynamics modeling has the problems of nonlinearity, uncertainty and the like which are currently difficult to solve, in the multi-body dynamics modeling method in the prior art, only a simple iterative correction is performed from the whole situation to replace manual trial and error, but an accurate multi-body dynamics model cannot be established for an automobile with more complex structure and function, so that the accuracy of the established multi-body dynamics model cannot meet the actual engineering requirements, and the accuracy, the applicability and the precision of the established multi-body dynamics model need to be improved.
Disclosure of Invention
The invention provides a method and a system for modeling automobile multi-body dynamics, which can improve the accuracy, the reliability and the applicability of the multi-body dynamics model establishment and improve the modeling efficiency.
In order to solve the above technical problem, an embodiment of the present invention provides an automotive multi-body dynamics modeling method, including
Establishing an initial multi-body dynamic model according to a vibration model and a Lagrange equation of the automobile;
according to load spectrum data of key points in a vibration transmission path collected by a road condition test and a first preset iteration ending condition, carrying out virtual iteration on the initial multi-body dynamic model, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirements of preset precision indexes according to a first iteration result and the precision indexes; the initial multi-body dynamic model consists of a plurality of local models; establishing a precision index according to a working condition transmission path analysis method and the transmission rate of a response measuring point in the vibration transmission path; the response measuring points are one or more selected from the key points;
if the initial multi-body dynamic model meets the requirements, taking the initial multi-body dynamic model as an optimal multi-body dynamic model;
if all the local models meet the requirements and the initial multi-body dynamic model does not meet the requirements, performing response surface optimization evaluation on the initial multi-body dynamic model according to a first iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model;
and if any local model does not meet the requirements, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result until all local models meet the preset precision index requirements, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model.
According to the embodiment of the invention, an initial multi-body dynamic model is established according to a vibration model and a Lagrange equation of an automobile, modeling is carried out by combining the Lagrange equation and the vibration model, a precision index is established by means of a working condition vibration transfer rate method, iterative checking is carried out on a local model by combining load spectrum data, precision evaluation is carried out on a complete model, overall to local index decomposition is completed, reliability precision indexes are met, the initial multi-body dynamic model is used as a final model, the efficiency of model establishment is greatly improved, iteration is carried out on the local model which does not meet the precision requirement, response surface optimization evaluation is carried out through an iteration result and an orthogonal Latin hypercube test, an optimal multi-body dynamic model meeting the reliability precision indexes is established, and the accuracy, reliability and applicability of modeling are effectively improved.
As a preferred scheme, an initial multi-body dynamic model is established according to a vibration model and a Lagrange equation of an automobile, and the method specifically comprises the following steps:
performing mass damping abstraction according to the structural combination, vibration excitation and response of the automobile system, and establishing a vibration model;
and establishing a dynamic equation according to the Lagrange equation and the vibration model, substituting the component parameters of the vibration model into the dynamic equation for calculation, and establishing an initial multi-body dynamic model.
As an optimal scheme, the accuracy index is established according to a working condition transmission path analysis method and the transmission rate of a response measuring point in the vibration transmission path, and specifically comprises the following steps:
establishing a transfer equation according to a working condition transfer path analysis method, calculating a first transfer rate matrix according to the transfer equation, and establishing a precision index according to the first transfer rate matrix and the transfer rate of a response measuring point in the vibration transfer path
As an optimal scheme, according to load spectrum data of key points in a vibration transmission path collected by a road condition test and a first preset iteration ending condition, performing virtual iteration on an initial multi-body dynamic model, and according to a first iteration result and a precision index, respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index, specifically:
arranging a sensing device at a key point of a vibration transmission path in the vibration model, and carrying out road condition test response measurement to obtain load spectrum data;
selecting an output point of a component in a vibration transmission path;
and performing virtual iteration according to the load spectrum data and a Newton-Raphson formula as follows:
u k+1 (s)＝u k (s)+f -1 (s)(y d (s)-y k (s))；
k＝1,2,3,…,n-1；
wherein u is k (s) is the kth excitation signal, y k (s) response signal, u, for the k-th excitation signal iteration 1 (s) is the initial excitation signal, f -1 (s) is a reverse transfer function, y d (s) load spectrum data;
when a first preset iteration ending condition is met, stopping virtual iteration;
inputting the excitation signal of the kth time into an initial multi-body dynamic model to obtain a model response signal of an output point; the excitation signal of the kth time is an excitation signal meeting a first preset iteration ending condition;
establishing a second transmissibility matrix according to the model response signal;
and respectively judging whether each local model and the initial multi-body dynamics model meet the requirements of preset precision indexes according to the first transfer rate matrix, the second transfer rate matrix and the precision indexes.
By implementing the embodiment of the invention, only the response signal of the road condition test is needed in the whole evaluation process of the reliability accuracy index, so that the experiment cost is greatly reduced, and the experiment period is shortened.
As a preferred scheme, if any local model does not meet the requirements, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result until all local models meet the preset precision index requirements, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal latin hypercube test, and establishing an optimal multi-body dynamic model, specifically:
if any local model does not meet the requirements, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result and the Newton-Raphson formula, wherein the virtual iteration is as follows:
u l+1 (s)＝u l (s)+f -1 (s)(y b (s)-y l (s))；
l＝1,2,3,…,n-1；
wherein u is l (s) is the kth excitation signal, y l (s) response signal, u, for the k-th excitation signal iteration 1 (s) is the initial excitation signal, f -1 (s) is a reverse transfer function, y b (s) is the result of the first iteration;
modifying a preset iteration ending condition according to the relative error value of the response curve, and stopping virtual iteration when the modified iteration ending condition is met;
inputting the excitation signal of the first time into the initial multi-body dynamic model to obtain a modified model response signal of an output point; wherein, the excitation signal of the first time is the excitation signal meeting the modified iteration end condition;
according to the modified model response signal, establishing a modified transfer rate matrix to obtain a modified iteration result;
judging whether each local model meets the requirement of a preset precision index or not according to the first transfer rate matrix, the modified transfer rate matrix and the precision index;
and if any local model does not meet the requirements, performing virtual iteration on the initial multi-body dynamic model again according to the modified iteration result until all local models meet the preset precision index requirements to obtain a second iteration result, and performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test to establish an optimal multi-body dynamic model.
By implementing the embodiment of the invention, if any local model does not meet the requirements, the iteration ending condition is further modified according to the relative error value of the response curve, and the initial multi-body dynamic model is continuously and virtually iterated, so that the local model meets the precision index, and the element to be corrected can be found out according to the iteration result, thereby being beneficial to further optimizing the global model (the initial multi-body dynamic model).
As a preferred scheme, according to an iteration result and a preset orthogonal latin hypercube test, response surface optimization evaluation is performed on an initial multi-body dynamic model, and an optimal multi-body dynamic model is established, wherein the iteration result is a first iteration result or a second iteration result, and specifically comprises the following steps:
determining an element to be corrected according to an iteration result, and analyzing sensitivity and contribution degree according to the element to be corrected to obtain a design variable; establishing a first design variable matrix according to the design variables and a preset first orthogonal Latin hypercube test;
and performing response surface optimization evaluation on the initial multi-body dynamic model according to the first design variable matrix, the design variables and the precision index, and establishing an optimal multi-body dynamic model.
As an optimal scheme, according to a first design variable matrix, design variables and precision indexes, response surface optimization evaluation is performed on an initial multi-body dynamic model, and an optimal multi-body dynamic model is established, specifically:
combining the first design variable matrix with the initial multi-body dynamic model to obtain a first response;
establishing a first response surface model according to the design variable value, the first response value and the response surface approximation function;
performing precision verification on the first response surface model, and evaluating whether the first response surface model meets the preset response surface precision requirement;
if so, establishing an optimized response surface model according to the design variables, the first response and the convolutional neural network, and establishing an optimal multi-body dynamic model according to the optimized response surface model;
if not, presetting a second orthogonal Latin hypercube test, establishing a second response surface model according to the second orthogonal Latin hypercube test and design variables until the second response surface model meets the preset response surface accuracy requirement, stopping the design of the orthogonal Latin hypercube test, and entering the subsequent steps that the response surface model meets the preset response surface accuracy requirement.
By implementing the embodiment of the invention, the response surface model is subjected to precision verification, the model with the highest similarity between the dynamic model and the response surface model is found out, the response surface model meeting the precision is subjected to convolutional neural network optimization, the optimal multi-body dynamic model is established, and the precision and the accuracy of the multi-body dynamic model are effectively improved.
As a preferred scheme, a second orthogonal latin hypercube test is preset, a second response surface model is established according to the second orthogonal latin hypercube test and design variables, until the second response surface model meets the preset response surface accuracy requirement, the design of the orthogonal latin hypercube test is stopped, and the following steps that the response surface model meets the preset response surface accuracy requirement are carried out, specifically:
presetting a second orthogonal Latin hypercube test, and establishing a second design variable matrix according to the second orthogonal Latin hypercube test and design variables;
combining the second design variable matrix with the initial multi-body dynamic model to obtain a second response;
establishing a second response surface model by using the design variable value, the second response value and the response surface approximation function;
performing precision verification on the second response surface model, and evaluating whether the second response surface model meets the preset response surface precision requirement;
if the second response surface model does not meet the preset response surface precision requirement, modifying the preset orthogonal Latin hypercube test to obtain a third response, carrying out precision verification on the third response surface model until the third response surface model meets the preset response surface precision requirement, and stopping the design of the orthogonal Latin hypercube test;
entering a subsequent step that the response surface model meets the preset response surface precision requirement, establishing an optimized response surface model according to the design variable, the modified response and the convolutional neural network, and establishing an optimal multi-body dynamic model according to the optimized response surface model; wherein the modified response is either the second response or the third response.
By implementing the embodiment of the invention, the response surface model is continuously corrected by designing the second orthogonal Latin hypercube test for multiple times, and the response surface model with higher precision is established, so that a foundation is laid for establishing an accurate multi-body dynamics model.
As a preferred scheme, an optimized response surface model is established according to design variables, responses and a convolutional neural network, and an optimal multi-body dynamic model is established according to the optimized response surface model, wherein the response is a first response or a modified response, and specifically comprises the following steps:
training a convolutional neural network according to design variables and responses, establishing an optimized response surface model according to a convolutional neural network training result, establishing an optimal multi-body dynamic model according to the optimized response surface model until the optimal multi-body dynamic model meets the preset precision index requirement, stopping the training of the convolutional neural network to obtain the optimal response surface model, and establishing the optimal multi-body dynamic model according to the optimal response surface model.
By implementing the embodiment of the invention, the response surface model is continuously optimized by training the convolutional neural network, and the training is stopped when the precision requirement of the precision response surface is met, so that the optimal response surface model is obtained, and the accurate model establishment and correction are carried out.
In order to solve the same technical problem, an embodiment of the present invention further provides an automotive multi-body dynamics modeling system, including: the device comprises an initial module, a precision index module, an evaluation module, a precision meeting module, a local optimization module and a global optimization module;
the initial module is used for establishing an initial multi-body dynamic model according to a vibration model and a Lagrange equation of the automobile, and the initial multi-body dynamic model consists of a plurality of local models;
the precision index module is used for establishing a precision index according to a working condition transmission path analysis method and the transmission rate at a response measurement point in the vibration transmission path; the response measuring points are one or more selected from the key points;
the evaluation module is used for performing virtual iteration on the initial multi-body dynamic model according to load spectrum data of key points in the vibration transmission path collected by the road condition test and a first preset iteration ending condition, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirements of preset precision indexes or not according to a first iteration result and the precision indexes;
the accuracy meeting module is used for taking the initial multi-body dynamic model as an optimal multi-body dynamic model if each local model and each initial multi-body dynamic model meet the requirement of accuracy indexes;
the local optimization module is used for performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result if any local model does not meet the requirement of the precision index until all local models meet the preset precision index requirement, and then obtaining a second iteration result;
and the global optimization module is used for carrying out response surface optimization evaluation on the initial multi-body dynamic model according to an iteration result and a preset orthogonal Latin hypercube test and establishing an optimal multi-body dynamic model if the initial multi-body dynamic model does not meet the requirement of the precision index.
Drawings
FIG. 1: the invention provides a flow diagram of one embodiment of the automobile multi-body dynamics modeling method;
FIG. 2: the simple flow chart for establishing the multi-body dynamics model of the embodiment of the automobile multi-body dynamics modeling method is provided by the invention;
FIG. 3: the invention provides a vibration model diagram of the multi-body dynamics of a truck in one embodiment of the multi-body dynamics modeling method of the truck;
FIG. 4: the invention provides a response surface optimization evaluation flow chart of one embodiment of the automobile multi-body dynamics modeling method;
FIG. 5 is a schematic view of: the invention provides a structural schematic diagram of an embodiment of an automobile multi-body dynamics modeling system.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
Referring to fig. 1, a schematic flow chart of a modeling method for multi-body dynamics of an automobile according to an embodiment of the present invention is shown, and fig. 2 is a simple flow chart of building a multi-body dynamics model. According to the modeling method, iterative checking, response surface optimization and multiple precision evaluation are carried out on the model by means of a working condition vibration transfer rate method, an optimal multi-body dynamics model is established, and the modeling precision, reliability and applicability are effectively improved. The modeling method comprises the following steps of 101 to 104, wherein the following steps are specifically carried out:
step 101: and establishing an initial multi-body dynamic model according to the vibration model of the automobile and the Lagrange equation.
Step 101 includes steps 1011 to 1012, each of which is as follows:
step 1011: performing mass damping abstraction according to the structural combination, vibration excitation and response of the automobile system, and establishing a vibration model;
in the present embodiment, the structural combination of the automobile research system is determined, then the vibration excitation and response of the research system are determined, finally the abstraction of mass block, damping and spring is performed according to the vibration transmission path of the system, and a schematic diagram of the vibration model is determined, taking a truck as an example, the truck multi-body dynamic vibration model is shown in fig. 3, and the relevant parameters in the truck vibration model are shown in table 1 below.
Table 1 cargo vehicle parameter table
Step 1012: and establishing a kinetic equation according to the Lagrange equation and the vibration model, substituting the component parameters of the vibration model into the kinetic equation for calculation, and establishing an initial multi-body kinetic model.
In this embodiment, the dynamical equation is determined according to the lagrangian equation and the vibration model of the belonging car, where the lagrangian equation is as follows:
in the formula, E k 、E p 、E d The kinetic energy, the potential energy and the dissipation energy of the system are obtained; q. q.s i 、
Generalized coordinates and generalized speed of the system; f i Is a reaction with q i Corresponding generalized force.
The dynamic model parameters are mass parameters, rigidity, damping and geometric dimensions of each component in the vibration model, so that specific parameters (mass, rotational inertia, inertia product, rigidity, damping and geometric parameters) can be further substituted into a dynamic equation for specific calculation, and an initial multi-body dynamic model is established.
Step 102: and establishing a precision index according to a working condition transmission path analysis method and the transmission rate of a response measuring point in the vibration transmission path.
Step 102 specifically comprises: establishing a transfer equation according to a working condition transfer path analysis method, calculating a first transfer rate matrix according to the transfer equation, and establishing a precision index according to the first transfer rate matrix and the transfer rate of a response measuring point in the vibration transfer path.
In the present embodiment, the OPTA equation (transfer equation) is established according to the operating condition transfer path analysis method as follows:
in the formula (I), the compound is shown in the specification,
for the response quantity of the target set of response points a,
for the response of the target response point set B, the superscript (i) represents the measured frequency domain signal of the ith group of data blocks, and the transfer rate matrix can be calculated by the following formula:
in the formula, the upper corner mark "+" represents the pseudo inverse of the matrix, V is unitary matrix of r × r dimension, U is unitary matrix of B × B dimension,
in order to obtain a B multiplied by r dimensional singular value matrix with all non-diagonal elements zero after removing smaller singular values, fc is the external force applied to the system.
Establishing a vibration transmissibility accuracy index, and determining transmissibility at two different response measurement points p and q at a frequency band [ f ] 1 ,f 2 ]The transmission rate precision indexes are as follows:
in the formula, D pq (f 1 ,f 2 ) As an index of the accuracy,
is a transmissibility matrix for a real vehicle test,
is the transmissibility matrix of the kinetic model (first transmissibility matrix).
Step 103: according to load spectrum data of key points in a vibration transmission path collected by a road condition test and a first preset iteration ending condition, carrying out virtual iteration on the initial multi-body dynamic model, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirements of preset precision indexes according to a first iteration result and the precision indexes; wherein the initial multi-body kinetic model consists of several local models.
In this embodiment, the local model is specifically an engine submodel, a cab submodel, a vehicle frame model, and the like of the initial multi-body dynamic vibration model, the global model is a multi-body dynamic model representing the whole, the final goal of building the dynamic model is to meet the requirement of the global model in the engineering, if the global model is met finally, the error of the local model is not affected, and if the local model is met locally and the global model also meets the preset accuracy index (reliability index), the model is directly built without subsequent optimization, and the initial multi-body dynamic model is used as the optimal multi-body dynamic model. The local model can not satisfy the basic error requirement of the engineering, and can not ignore the local realistic constraint condition, so if each local model can not satisfy the requirement, iterative check is needed to find the variable to be modified, and design optimization is carried out on the global model, or when the local model can satisfy the precision requirement and the global model can not satisfy the precision requirement, design optimization is also needed to be carried out on the global model.
Step 103 specifically comprises: arranging a sensing device at a key point of a vibration transmission path in the vibration model, and carrying out road condition test response measurement to obtain load spectrum data;
selecting an output point of a component in a vibration transmission path;
and performing virtual iteration according to the load spectrum data and a Newton-Raphson formula as follows:
u k+1 (s)＝u k (s)+f -1 (s)(y d (s)-y k (s))；
k＝1,2,3,…,n-1；
wherein u is k (s) is the kth excitation signal, y k (s) response signal, u, for the k-th excitation signal iteration 1 (s) is the initial excitation signal(s),f -1 (s) is a reverse transfer function, y d (s) load spectrum data;
stopping virtual iteration when a first preset iteration ending condition is met;
inputting the excitation signal of the kth time into an initial multi-body dynamic model to obtain a model response signal of an output point; the excitation signal of the kth time is an excitation signal meeting a first preset iteration ending condition;
establishing a second transmissibility matrix according to the model response signal;
and respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index or not according to the first transfer rate matrix, the second transfer rate matrix and the precision index.
In this embodiment, a sensor is arranged at a key point of a vibration transmission path in a vibration model, a road condition test response measurement is performed, load spectrum data of a test is obtained, the load spectrum data is a response signal, and the key point of the response is arranged at a cab seat, a joint of a cab suspension and a frame, a joint of the frame and a front suspension, a joint of the frame and a rear suspension, a joint of an engine and the frame, and a joint of a tire and an axle, for example, a truck.
Performing iterative checking according to the load spectrum data and a preliminarily established dynamic model, selecting output points of components on a transmission path (the output points of the components comprise a reference output point and a non-reference output point), wherein the reference output point and the non-reference output point are corresponding points of an actual object of the vibration model, selecting any component arranged on the vibration transmission path, generally establishing a transmission path matrix of test data at the connecting points of any 2 components, and in the dynamic model, iterating the input of the system according to a Newton-Raphson formula:
u k+1 (s)＝u k (s)+f -1 (s)(y d (s)-y k (s))
k＝1,2,3,…,n-1
in the formula, y k (s) is the k-th excitation signal u k (s) the iterated response signal, u 1 (s) is the initial excitation signal, f -1 (s) isInverse transfer function, y d (s) is the load spectrum data (measured response signal).
When the response curves of the response signal under the iterative excitation and the response signal under the experiment in the time domain and the frequency domain are basically consistent, and the relative error RMS value is smaller than a preset value (such as 0.2), namely a first preset iteration ending condition is met, stopping virtual iteration;
and determining the response of a reference output point and the response of a non-reference output point by the excitation input kinetic model, establishing a transfer rate matrix of the kinetic model, and evaluating whether the accuracy of the preliminarily established model meets the requirement or not by referring to the accuracy index of the transfer rate. Inputting the excitation signal of the kth time into an initial multi-body dynamic model, and determining the response of a reference output point and the response of a non-reference output point, namely the model response signal of the output point; the excitation signal of the kth time is an excitation signal meeting a first preset iteration ending condition;
establishing a transfer rate matrix (a second transfer rate matrix) of the real vehicle test by the transfer rate matrix calculation method in the step 102 according to the model response signal of the output point;
calculating a precision index value according to the transfer rate matrix (first transfer rate matrix) of the dynamic model and the transfer rate matrix (second transfer rate matrix) tested by the real vehicle test by using the transfer rate precision index formula in the step 102, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index (preset credibility value) according to the comparison between the preset precision index value and the calculated precision index value.
Step 104: and if any local model does not meet the requirement, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result until all local models meet the preset precision index requirement, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model.
Step 104 specifically includes: if any local model does not meet the requirements, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result and the Newton-Raphson formula, wherein the virtual iteration is as follows:
u l+1 (s)＝u l (s)+f -1 (s)(y b (s)-y l (s))；
l＝1,2,3,…,n-1；
wherein u is l (s) is the kth excitation signal, y l (s) response signal, u, for the k-th excitation signal iteration 1 (s) is the initial excitation signal, f -1 (s) is a reverse transfer function, y b (s) is the result of the first iteration;
modifying a preset iteration ending condition according to the relative error value of the response curve, and stopping virtual iteration when the modified iteration ending condition is met;
inputting the excitation signal of the first time into the initial multi-body dynamic model to obtain a modified model response signal of an output point; wherein, the excitation signal of the first time is the excitation signal meeting the modified iteration end condition;
according to the modified model response signal, establishing a modified transfer rate matrix to obtain a modified iteration result;
judging whether each local model meets the requirement of a preset precision index or not according to the first transfer rate matrix, the modified transfer rate matrix and the precision index;
and if any local model does not meet the requirement, performing virtual iteration on the initial multi-body dynamic model again according to the modified iteration result until all local models meet the preset precision index requirement, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model.
In this embodiment, if any local model does not satisfy the requirements, according to the same virtual iteration method, the first iteration result is an initial response signal of a new iteration, and meanwhile, according to a specific value of a relative error value of a response curve in the last virtual iteration, an end condition of the new iteration is modified, for example, when the relative error RMS values of the response curves in the time domain and the frequency domain of the response signal under the excitation of the iteration and the response signal under the experiment are smaller than a preset value (for example, 0.1), that is, when the modified preset iteration end condition is satisfied, the virtual iteration is stopped, whether the local model satisfies the preset precision index requirement is judged according to the same method in step 103, until all the local models satisfy the preset precision index requirement, a second iteration result is obtained, then, the initial multi-body dynamics model (global model) is optimized, and according to the second iteration result and the preset orthogonal latin hypercube test, the response surface optimization evaluation is performed on the initial multi-body dynamics model, and the optimal multi-body dynamics model is established.
Step 105: if all the local models meet the requirements and the initial multi-body dynamic model does not meet the requirements, performing response surface optimization evaluation on the initial multi-body dynamic model according to a first iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model;
step 105 includes steps 1051 to 1053, each of which is as follows:
step 1051: determining an element to be corrected according to an iteration result, and analyzing sensitivity and contribution degree according to the element to be corrected to obtain a design variable;
in this embodiment, the iteration result is the first iteration result or the second iteration result according to the actual evaluation process, and the vibration transmission path and the component to be corrected are determined according to the iteration result. According to the sensitivity formula, the formula is as follows:
where x is the target point output response, k is the stiffness, c is the damping, and l is the number of components on the transmission path.
According to the response contribution formula, the formula is as follows:
in the formula (I), the compound is shown in the specification,
and
values representing local stiffness and damping changes by equal amounts;
and
representing the stiffness and damping of the original model;
and
respectively representing the values of the overall variation of the original model, x,
displacement and velocity responses, respectively.
And determining the design variables of model correction through a sensitivity formula, a contribution formula and an initial multi-body dynamic model (dynamic equation).
Step 1052: and establishing a first design variable matrix according to the design variables and a preset first orthogonal Latin hypercube test.
In this embodiment, random permutation of q independent 1,2, … …, p is taken, and then the permutations are respectively taken as column vectors to form a p × q design matrix, which is specifically shown in that 2 columns are orthogonal in the design, the square column of any column of elements and the dot product of any 2 columns of elements in the design are orthogonal in other columns of the design, the given experiment frequency is 2^ (c + 1) (c ≧ 1), and the factor number is 2^c, then the hyper latin square experiment construction method is LHD (2 ^ (c + 1), 2^c).
And forming a design variable matrix by using sample points determined by the orthogonal Latin hypercube test design according to the design variables and a preset orthogonal Latin hypercube test by adopting a hyper Latin square test construction method, and establishing a first design variable matrix if a first orthogonal Latin hypercube test is preset.
Step 1053: and performing response surface optimization evaluation on the initial multi-body dynamic model according to the first design variable matrix, the design variables and the precision index, and establishing an optimal multi-body dynamic model.
Step 1053 is specifically shown in the response surface optimization evaluation flow of fig. 4, and includes steps 401 to 404, and each step is specifically as follows:
step 401: combining the first design variable matrix with the initial multi-body dynamic model to obtain a first response;
in the embodiment, the design variable matrix is determined through an orthogonal Latin hypercube experimental design, the response is obtained by combining the design variable matrix with a kinetic equation, and the first response is obtained according to the first design variable matrix.
Step 402: establishing a first response surface model according to the design variable value, the first response value and the response surface approximation function;
in this embodiment, an incomplete 4 th order polynomial is used as the response surface approximation function, and the formula is as follows:
in the formula (I), the compound is shown in the specification,
as response surface approximation function, x i Design variables, beta, for the model (0,i.ij,ii,iii,iiii) Is the undetermined coefficient.
Substituting the design variable value and the response value into the response surface approximation function to determine the undetermined coefficient beta (0,i.ij,ii,iii,iiii) Thus, a response surface model, i.e. a response surface approximation function of the determined coefficients, is established. According to the design variable value, the first response value andand establishing a first response surface model by using the response surface approximation function.
Step 403: performing precision verification on the first response surface model, and evaluating whether the first response surface model meets the preset response surface precision requirement or not;
in this embodiment, the accuracy of the established response surface model is verified, and the root mean square residual value RMSE and the decision coefficient R are used 2 Verification is performed, and the formula is as follows:
in the formula, k is the number of samples;
calculating a value of y for the response surface i Is calculated for the initial multi-body dynamic model,
the mean of the results was calculated for the initial multi-body kinetic model. When the RMSE tends to 0, the response surface error is small; when R is 2 When the similarity of the response surface and the dynamic model is close to 1, the similarity is high, and the RMSE value and R of the accuracy requirement of the response surface model are preset 2 Value, RMSE value and R calculated by verification 2 Comparing the value with a predetermined value, calculating that the RMSE value is less than the predetermined RMSE value, calculating R 2 A value greater than a predetermined R 2 And the value meets the requirement of a preset value, namely the response surface model meets the requirement of the preset response surface precision.
Step 404: and if the first response surface model does not meet the preset response surface precision requirement, presetting a second orthogonal Latin hypercube test, establishing a second response surface model according to the second orthogonal Latin hypercube test and the design variables until the second response surface model meets the preset response surface precision requirement, stopping the design of the orthogonal Latin hypercube test, and entering the subsequent step that the response surface model meets the preset response surface precision requirement.
Step 404 specifically includes: presetting a second orthogonal Latin hypercube test, and establishing a second design variable matrix according to the second orthogonal Latin hypercube test and design variables;
combining the second design variable matrix with the initial multi-body dynamic model to obtain a second response;
establishing a second response surface model by using the design variable value, the second response value and the response surface approximation function;
performing precision verification on the second response surface model, and evaluating whether the second response surface model meets the preset response surface precision requirement;
if the second response surface model does not meet the preset response surface precision requirement, modifying the preset orthogonal Latin hypercube test to obtain a third response, carrying out precision verification on the third response surface model until the third response surface model meets the preset response surface precision requirement, and stopping the design of the orthogonal Latin hypercube test;
entering a subsequent step that the response surface model meets the preset response surface precision requirement, establishing an optimized response surface model according to the design variable, the modified response and the convolutional neural network, and establishing an optimal multi-body dynamic model according to the optimized response surface model; wherein the modified response is the second response or the third response.
In this embodiment, when the accuracy of the first response surface model does not meet the requirement, the orthogonal latin hypercube test design needs to be modified continuously, the preset sample point is changed, the design variable matrix is changed, the second design variable matrix is obtained, after the second design variable matrix is obtained, the second response surface model is established according to the same method as that in step 402, whether the second response surface model meets the preset response surface accuracy requirement is judged according to the same method as that in step 403, if the second response surface model still does not meet the preset response surface accuracy requirement, the preset orthogonal latin hypercube test is modified continuously, the accuracy verification of the modified response surface model (third response surface model) is performed, the orthogonal latin hypercube test design and the accuracy verification of the response surface model are modified continuously until the third response surface model meets the preset response surface accuracy requirement, a modified response (second response or third response) and the modified design variable matrix can be obtained therefrom, the orthogonal latin hypercube test design is stopped, and the response surface model continues to enter a subsequent step where the response surface model meets the preset response surface accuracy requirement, that the neural network accuracy requirement is satisfied next.
Step 405: if the response surface model meets the preset response surface precision requirement, establishing an optimized response surface model according to the design variables, the response and the convolutional neural network, and establishing an optimal multi-body dynamic model according to the optimized response surface model, wherein the response surface model comprises a first response surface model, a second response surface model and a modified response surface model; the response is either a first response or a modified response.
Step 405 specifically includes: training a convolutional neural network according to design variables and responses, establishing an optimized response surface model according to a training result of the convolutional neural network, establishing an optimized multi-body dynamic model according to the optimized response surface model until the optimized multi-body dynamic model meets the requirement of a preset precision index, stopping training of the convolutional neural network to obtain an optimal response surface model, and establishing an optimal multi-body dynamic model according to the optimal response surface model.
In this embodiment, the calculation formula according to the l-th layer convolutional neural network is as follows:
g(.)＝max(0,x)
in the formula (I), the compound is shown in the specification,
an nth feature representing an output value of the l-th layer,
a weight matrix representing the nth convolution kernel of the l-th layer,
the output of the l-1 th layer is shown,
representing a paranoia term, and g (.) representing an activation function.
As an example of neural network design, the initial value of the number of network layers is 2 layers, the number of convolution kernels is 12, the size is 3, the pooling layers are all maximum pooling, the size is 2, the batch size is 300, epoch =2000, the learning rate is 0.01, dropout =0.5, the simulation scenario is set to SNR =20dB, and the sequence length is 10s.
Training a convolutional neural network according to input (design variable) of an orthogonal Latin hypercube test and output response (response is obtained by combining a design variable matrix with an initial multi-body dynamics equation) of a response surface model, wherein the design variable matrix is the design variable matrix when the response surface model meets the accuracy requirement, and can be a first design variable matrix, a second design variable matrix or a third design variable matrix and other design variable matrices, the response is the response when the response surface model meets the accuracy requirement, and can be the response after the first response or modification, namely the response such as the first response, the second response or the third response, and the like, training the convolutional neural network, putting the optimized design variable value into the initial multi-body dynamics model to obtain an optimized multi-body dynamics model, verifying whether the optimized multi-body dynamics model meets the requirement of a preset accuracy index again, continuously training the neural network until the requirement is met by precision evaluation of the optimized multi-body dynamics model, obtaining the optimized response surface model combined with a neural network algorithm, and finishing the establishment of the optimal multi-body dynamics model.
Step 106: if the initial multi-body dynamic model meets the requirements, taking the initial multi-body dynamic model as an optimal multi-body dynamic model;
in this embodiment, when it is determined that each local model and the initial multi-body kinetic model both meet the requirement of the accuracy index, the multi-body kinetic model is directly established without subsequent optimization, the steps of multiple iterative optimization are reduced, the modeling efficiency is improved, and the initial multi-body kinetic model is used as the optimal multi-body kinetic model.
According to the embodiment of the invention, an initial multi-body dynamic model is established according to a vibration model and a Lagrange equation of an automobile, modeling is carried out by combining the Lagrange equation and the vibration model, a precision index is established by means of a working condition vibration transfer rate method, iterative checking is carried out on a local model by combining load spectrum data, precision evaluation is carried out on a complete model, overall to local index decomposition is completed, reliability precision indexes are met, the initial multi-body dynamic model is used as a final model, the efficiency of model establishment is greatly improved, iteration is carried out on the local model which does not meet the precision requirement, response surface optimization evaluation is carried out through an iteration result and an orthogonal Latin hypercube test, an optimal multi-body dynamic model meeting the reliability precision indexes is established, and the accuracy, reliability and applicability of modeling are effectively improved.
Example two
Correspondingly, referring to fig. 5, fig. 5 is a schematic structural diagram of a second embodiment of the automotive multi-body dynamics modeling system provided by the present invention. As shown in fig. 5, the automotive multi-body dynamics modeling system includes: an initial module 501, a precision index module 502, an evaluation module 503, a satisfaction precision module 504, a local optimization module 505, and a global optimization module 506.
The initial module 501 is configured to establish an initial multi-body dynamic model according to a vibration model of an automobile and a lagrangian equation, where the initial multi-body dynamic model is composed of a plurality of local models;
the precision index module 502 is used for establishing a precision index according to a working condition transmission path analysis method and the transmission rate at a response measurement point in the vibration transmission path; the response measuring points are one or more selected from the key points;
the evaluation module 503 is configured to perform virtual iteration on the initial multi-body dynamic model according to load spectrum data of key points in the vibration transmission path collected by the road condition test and a first preset iteration ending condition, and respectively determine whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index according to a first iteration result and the precision index;
the meet-precision module 504 is configured to take the initial multi-body kinetic model as the optimal multi-body kinetic model if each local model and the initial multi-body kinetic model meet the requirement of the precision index;
the local optimization module 505 is configured to, if any local model does not meet the requirement of the accuracy index, perform virtual iteration on the initial multi-body dynamic model again according to the first iteration result until all local models meet the preset accuracy index requirement, and obtain a second iteration result;
the global optimization module 506 is configured to perform response surface optimization evaluation on the initial multi-body kinetic model according to an iteration result and a preset orthogonal latin hypercube test if the initial multi-body kinetic model does not meet the requirement of the precision index, and establish an optimal multi-body kinetic model.
According to the embodiment of the invention, an initial multi-body dynamic model is established according to a vibration model and a Lagrange equation of an automobile, modeling is carried out by combining the Lagrange equation and the vibration model, a precision index is established by means of a working condition vibration transfer rate method, iterative checking is carried out on a local model by combining load spectrum data, precision evaluation is carried out on a complete model, overall to local index decomposition is completed, reliability precision indexes are met, the initial multi-body dynamic model is used as a final model, the efficiency of model establishment is greatly improved, iteration is carried out on the local model which does not meet the precision requirement, response surface optimization evaluation is carried out through an iteration result and an orthogonal Latin hypercube test, an optimal multi-body dynamic model meeting the reliability precision indexes is established, and the accuracy, reliability and applicability of modeling are effectively improved.
The above-mentioned embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, and it should be understood that the above-mentioned embodiments are only examples of the present invention and are not intended to limit the scope of the present invention. It should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. An automotive multi-body dynamics modeling method, comprising:
establishing an initial multi-body dynamic model according to a vibration model and a Lagrange equation of the automobile;
performing virtual iteration on the initial multi-body dynamic model according to load spectrum data of key points in a vibration transmission path acquired by a road condition test and a first preset iteration ending condition, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index or not according to a first iteration result and the precision index; wherein the initial multi-body kinetic model consists of a number of local models; the precision index is established according to a working condition transmission path analysis method and the transmission rate of a response measuring point in the vibration transmission path; the response measuring point is one or more selected from the key points;
if the initial multi-body dynamic model meets the requirements, taking the initial multi-body dynamic model as an optimal multi-body dynamic model;
if all the local models meet the requirements and the initial multi-body dynamic model does not meet the requirements, performing response surface optimization evaluation on the initial multi-body dynamic model according to the first iteration result and a preset orthogonal Latin hypercube test to establish an optimal multi-body dynamic model;
and if any local model does not meet the requirement, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result until all the local models meet the preset precision index requirement, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model.
2. The automotive multi-body dynamics modeling method of claim 1, wherein the initial multi-body dynamics model is established according to a vibration model of the automobile and a lagrangian equation, specifically:
performing mass damping abstraction according to the structural combination, vibration excitation and response of the automobile system, and establishing the vibration model;
and establishing a kinetic equation according to the Lagrange equation and the vibration model, substituting the component parameters of the vibration model into the kinetic equation for calculation, and establishing the initial multi-body kinetic model.
3. The automotive multi-body dynamics modeling method of claim 1, wherein the accuracy index is established based on a working condition transfer path analysis method and a transfer rate at a response measurement point in a vibration transfer path, specifically:
establishing a transfer equation according to a working condition transfer path analysis method, calculating a first transfer rate matrix according to the transfer equation, and establishing the precision index according to the first transfer rate matrix and the transfer rate of a response measuring point in the vibration transfer path.
4. The modeling method of automotive multi-body dynamics according to claim 3, wherein the virtual iteration is performed on the initial multi-body dynamics model according to the load spectrum data of the key points in the vibration transmission path collected by the road condition test and the first preset iteration ending condition, and whether each local model and the initial multi-body dynamics model meet the requirements of the preset accuracy index or not is respectively judged according to the first iteration result and the accuracy index, specifically:
arranging a sensing device at a key point of a vibration transmission path in the vibration model, and carrying out road condition test response measurement to obtain the load spectrum data;
selecting an output point of a component in the vibration transmission path;
performing the virtual iteration according to the load spectrum data and a Newton-Raphson formula as follows:
u k+1 (s)＝u k (s)+f -1 (s)(y d (s)-y k (s))；
k＝1,2,3,....,n-1；
wherein u is k (s) is the kth excitation signal, y k (s) response signal, u, for the k-th excitation signal iteration 1 (s) is the initial excitation signal, f -1 (s) is a reverse transfer function, y d (s) the load spectrum data;
stopping the virtual iteration when a first preset iteration ending condition is met;
inputting the excitation signal of the kth time into the initial multi-body dynamic model to obtain a model response signal of the output point; wherein the excitation signal of the kth time is an excitation signal meeting a first preset iteration end condition;
establishing a second transmissibility matrix according to the model response signal;
and respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of the preset precision index or not according to the first transfer rate matrix, the second transfer rate matrix and the precision index.
5. The automotive multi-body dynamics modeling method according to claim 4, wherein if any one of the local models does not meet the requirements, virtual iteration is performed on the initial multi-body dynamics model again according to the first iteration result until all the local models meet the preset accuracy index requirements, a second iteration result is obtained, response surface optimization evaluation is performed on the initial multi-body dynamics model according to the second iteration result and a preset orthogonal latin hypercube test, and an optimal multi-body dynamics model is established, specifically:
if any local model does not meet the requirement, performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result and the Newton-Raphson formula, wherein the formula is as follows:
u l+1 (s)＝u l (s)+f -1 (s)(y b (s)-y l (s))；
l＝1,2,3,....,n-1；
wherein u is l (s) is the kth excitation signal, y l (s) is the k-th excitation signalIterative response signal u 1 (s) is the initial excitation signal, f -1 (s) is a reverse transfer function, y b (s) is the first iteration result;
modifying a preset iteration ending condition according to the relative error value of the response curve, and stopping the virtual iteration when the modified iteration ending condition is met;
inputting the excitation signal of the first time into the initial multi-body dynamic model to obtain a modified model response signal of the output point; wherein the excitation signal of the first time is the excitation signal meeting the modified iteration end condition;
establishing a modified transmissibility matrix according to the modified model response signal to obtain a modified iteration result;
judging whether each local model meets the requirement of a preset precision index or not according to the first transfer rate matrix, the modified transfer rate matrix and the precision index;
and if any local model does not meet the requirement, performing virtual iteration on the initial multi-body dynamic model again according to the modified iteration result until all the local models meet the preset precision index requirement, obtaining a second iteration result, performing response surface optimization evaluation on the initial multi-body dynamic model according to the second iteration result and a preset orthogonal Latin hypercube test, and establishing an optimal multi-body dynamic model.
6. The automotive multi-body dynamics modeling method of claim 1, wherein the initial multi-body dynamics model is subjected to response surface optimization evaluation according to an iteration result and a preset orthogonal latin hypercube test to establish an optimal multi-body dynamics model, wherein the iteration result is the first iteration result or the second iteration result, specifically:
determining an element to be corrected according to the iteration result, and analyzing the sensitivity and the contribution degree according to the element to be corrected to obtain a design variable; establishing a first design variable matrix according to the design variables and a preset first orthogonal Latin hypercube test;
and performing response surface optimization evaluation on the initial multi-body dynamic model according to the first design variable matrix, the design variables and the precision index, and establishing an optimal multi-body dynamic model.
7. The modeling method of automotive multi-body dynamics according to claim 6, wherein the initial multi-body dynamics model is subjected to response surface optimization evaluation according to the first design variable matrix, the design variables and the accuracy index, and an optimal multi-body dynamics model is established, specifically:
combining the first design variable matrix with the initial multi-body kinetic model to obtain a first response;
establishing a first response surface model according to the design variable value, the first response value and the response surface approximation function;
performing precision verification on the first response surface model, and evaluating whether the first response surface model meets the precision requirement of a preset response surface;
if so, establishing an optimized response surface model according to the design variables, the first response and the convolutional neural network, and establishing the optimal multi-body dynamics model according to the optimized response surface model;
if not, presetting a second orthogonal Latin hypercube test, establishing a second response surface model according to the second orthogonal Latin hypercube test and the design variables, stopping the design of the orthogonal Latin hypercube test until the second response surface model meets the preset response surface precision requirement, and entering the subsequent step that the response surface model meets the preset response surface precision requirement.
8. The automotive multi-body dynamics modeling method according to claim 7, wherein the second orthogonal latin hypercube test is preset, and a second response surface model is established according to the second orthogonal latin hypercube test and the design variables until the second response surface model meets the preset response surface accuracy requirement, the design of the orthogonal latin hypercube test is stopped, and the subsequent step of the response surface model meeting the preset response surface accuracy requirement is carried out, specifically:
presetting a second orthogonal Latin hypercube test, and establishing a second design variable matrix according to the second orthogonal Latin hypercube test and the design variables;
combining the second design variable matrix with the initial multi-body kinetic model to obtain a second response;
establishing a second response surface model by using the design variable value, the second response value and the response surface approximation function;
performing precision verification on the second response surface model, and evaluating whether the second response surface model meets the preset response surface precision requirement;
if the second response surface model does not meet the preset response surface precision requirement, modifying the preset orthogonal Latin hypercube test to obtain a third response, carrying out precision verification on the third response surface model until the third response surface model meets the preset response surface precision requirement, and stopping the design of the orthogonal Latin hypercube test;
entering a subsequent step that a response surface model meets the precision requirement of a preset response surface, establishing an optimized response surface model according to the design variable, the modified response and the convolutional neural network, and establishing the optimal multi-body dynamic model according to the optimized response surface model; wherein the modified response is either a second response or a third response.
9. The automotive multi-body dynamics modeling method according to claim 7 or 8, wherein the optimal multi-body dynamics model is established according to the optimal response surface model, wherein the response is the first response or the modified response, and specifically:
training a convolutional neural network according to the design variables and the response, establishing an optimized response surface model according to the training result of the convolutional neural network, establishing an optimal multi-body dynamic model according to the optimized response surface model, stopping the training of the convolutional neural network until the optimal multi-body dynamic model meets the requirement of a preset precision index, obtaining the optimal response surface model, and establishing the optimal multi-body dynamic model according to the optimal response surface model.
10. An automotive multi-body dynamics modeling system, comprising: the device comprises an initial module, a precision index module, an evaluation module, a precision meeting module, a local optimization module and a global optimization module;
the initial module is used for establishing an initial multi-body dynamic model according to a vibration model and a Lagrange equation of the automobile, and the initial multi-body dynamic model consists of a plurality of local models;
the precision index module is used for establishing a precision index according to a working condition transmission path analysis method and the transmission rate at a response measuring point in the vibration transmission path; the response measuring points are one or more selected from the key points;
the evaluation module is used for performing virtual iteration on the initial multi-body dynamic model according to load spectrum data of key points in a vibration transmission path acquired by a road condition test and a first preset iteration ending condition, and respectively judging whether each local model and the initial multi-body dynamic model meet the requirement of a preset precision index according to a first iteration result and a precision index;
the accuracy satisfying module is used for taking the initial multi-body dynamic model as the optimal multi-body dynamic model if each local model and the initial multi-body dynamic model meet the requirement of the accuracy index;
the local optimization module is used for performing virtual iteration on the initial multi-body dynamic model again according to the first iteration result if any local model does not meet the requirement of the precision index until all local models meet the preset precision index requirement, and then obtaining a second iteration result;
and the global optimization module is used for carrying out response surface optimization evaluation on the initial multi-body dynamic model according to an iteration result and a preset orthogonal Latin hypercube test and establishing an optimal multi-body dynamic model if the initial multi-body dynamic model does not meet the requirement of the precision index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211116217.XA CN115577615A (en) | 2022-09-14 | 2022-09-14 | Automobile multi-body dynamics modeling method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211116217.XA CN115577615A (en) | 2022-09-14 | 2022-09-14 | Automobile multi-body dynamics modeling method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115577615A true CN115577615A (en) | 2023-01-06 |
Family
ID=84580647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211116217.XA Pending CN115577615A (en) | 2022-09-14 | 2022-09-14 | Automobile multi-body dynamics modeling method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115577615A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910940A (en) * | 2023-09-11 | 2023-10-20 | 山东大学 | Engineering machinery fatigue life analysis method, system, equipment and storage medium |
-
2022
- 2022-09-14 CN CN202211116217.XA patent/CN115577615A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910940A (en) * | 2023-09-11 | 2023-10-20 | 山东大学 | Engineering machinery fatigue life analysis method, system, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6704664B2 (en) | Fatigue sensitivity determination procedure | |
| Kong et al. | Optimization of spring fatigue life prediction model for vehicle ride using hybrid multi-layer perceptron artificial neural networks | |
| CN109033643B (en) | Sensitivity analysis-based automobile operation stability transient performance parameter optimization method | |
| CN116090111B (en) | Automobile leaf spring fatigue life prediction method based on deep learning model | |
| CN115577615A (en) | Automobile multi-body dynamics modeling method and system | |
| CN114239149B (en) | Commercial vehicle cab white body light weight method based on collision performance optimization | |
| CN113886974B (en) | Method for predicting sound path noise of in-vehicle structure | |
| CN115270296A (en) | Method and system for analyzing fatigue durability of commercial vehicle cab | |
| Luegmair et al. | Challenges in vibroacoustic vehicle body simulation including uncertainties | |
| CN115688288A (en) | Aircraft pneumatic parameter identification method and device, computer equipment and storage medium | |
| CN111797471A (en) | Engine hood lightweight design method based on radial basis function neural network approximate model | |
| CN113515808A (en) | Noise prediction method for power assembly | |
| Fu et al. | Prediction of automobile aerodynamic drag coefficient for SUV cars based on a novel XGBoost model | |
| CN114676648B (en) | Vehicle load spectrum prediction method and device based on machine learning | |
| Kanchwala et al. | A generalized quarter car modelling approach with frame flexibility and other nonlocal effects | |
| Horvath | Structural and System Models | |
| CN112560180B (en) | Transmission path analysis method of multipoint installation structure | |
| Van der Auweraer | Requirements and opportunities for structural testing in view of hybrid and virtual modelling | |
| Gawande et al. | Recreation of service loads using strain gauges of speed shifting lever | |
| Kjellberg et al. | Real-time nonlinear model predictive control for semi-active suspension with road preview | |
| CN116629078B (en) | Method and system for predicting fatigue life durability of automobile | |
| Van der Auweraer et al. | New approaches enabling NVH analysis to lead design in body development | |
| CN114647897B (en) | Cab structure optimization method based on hierarchical entropy-gray correlation method | |
| Metallidis et al. | Parametric identification and health monitoring of complex ground vehicle models | |
| CN114048548B (en) | Method for estimating dynamic load of air spring of heavy-duty vehicle |
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 |