CN116839783A - Method for measuring stress value and deformation of automobile leaf spring based on machine learning - Google Patents
Method for measuring stress value and deformation of automobile leaf spring based on machine learning Download PDFInfo
- Publication number
- CN116839783A CN116839783A CN202311118723.7A CN202311118723A CN116839783A CN 116839783 A CN116839783 A CN 116839783A CN 202311118723 A CN202311118723 A CN 202311118723A CN 116839783 A CN116839783 A CN 116839783A
- Authority
- CN
- China
- Prior art keywords
- test
- plate spring
- load
- neural network
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000010801 machine learning Methods 0.000 title claims abstract description 21
- 238000012360 testing method Methods 0.000 claims abstract description 212
- 238000003062 neural network model Methods 0.000 claims abstract description 73
- 238000012549 training Methods 0.000 claims abstract description 49
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000010606 normalization Methods 0.000 claims abstract description 10
- 238000006073 displacement reaction Methods 0.000 claims description 47
- 230000006870 function Effects 0.000 claims description 15
- 238000012795 verification Methods 0.000 claims description 13
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000003068 static effect Effects 0.000 claims description 5
- 230000006835 compression Effects 0.000 claims description 4
- 238000007906 compression Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 11
- 238000004088 simulation Methods 0.000 description 7
- 230000000875 corresponding effect Effects 0.000 description 6
- 241000205572 Jeffersonia diphylla Species 0.000 description 4
- 230000008901 benefit Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000008672 reprogramming Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01L—MEASURING FORCE, STRESS, TORQUE, WORK, MECHANICAL POWER, MECHANICAL EFFICIENCY, OR FLUID PRESSURE
- G01L5/00—Apparatus for, or methods of, measuring force, work, mechanical power, or torque, specially adapted for specific purposes
- G01L5/0057—Apparatus for, or methods of, measuring force, work, mechanical power, or torque, specially adapted for specific purposes measuring forces due to spring-shaped elements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B21/00—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant
- G01B21/32—Measuring arrangements or details thereof, where the measuring technique is not covered by the other groups of this subclass, unspecified or not relevant for measuring the deformation in a solid
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
-
- 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
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Computer Hardware Design (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Geometry (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
Abstract
The application provides a method for measuring stress value and deformation of an automobile leaf spring based on machine learning, which comprises the following steps: acquiring the full-load vertical load and the maximum longitudinal load of the test leaf spring; constructing a digital twin plate spring model of the test plate spring, and determining a patch position for pasting a strain gauge based on the digital twin plate spring model; bench test is carried out on the test plate spring pasted with the strain gauge so as to obtain a standby set; constructing an initial neural network model, carrying out normalization processing on the standby set, and training the initial neural network model through the normalized standby set to obtain a final neural network model; and pasting the strain gauge to the patch position of the plate spring to be tested, and carrying out road load test on the whole vehicle provided with the plate spring to be tested to obtain an actual strain value, and further determining the stress value and the deformation of the plate spring to be tested through a final neural network model. The method avoids the error generated in the measuring process of the sensor without using the sensor, and leads the stress value and the deformation to be more accurate.
Description
Technical Field
The application relates to the field of data processing, in particular to a method for measuring stress value and deformation of an automobile leaf spring based on machine learning.
Background
The durability is taken as one of important performance indexes of automobile products, directly relates to the safety and reliability of the operation of the automobile under various working conditions, and has important influence on the whole automobile quality and the reputation of an automobile enterprise. In the development stage of automobile products, the simulation technology based on digital twinning is required to carry out the simulation analysis of the durability life of the whole automobile parts and the whole automobile system, and the simulation solution analysis of the durability life is required to be carried out based on the road durability load born by each part. The road endurance load of the parts is required to be measured based on a real vehicle to obtain six component forces of the vehicle at the wheel center of the vehicle, then the six component forces are applied to the digital virtual model ADAMS digital twin model, then the load of each part is solved and calculated, and finally the fatigue endurance life analysis of the parts is carried out.
Because the whole-vehicle digital twin multi-body dynamics ADAMS model contains a large number of nonlinear rubber connecting pieces such as bushings and the like, the digital virtual prototype cannot be completely consistent with a physical sample vehicle, in order to ensure the accuracy of decomposed load, the actual deformation and actual stress value of a steel plate spring and the simulated deformation and simulated stress value calculated by a simulation model are usually required to be compared to ensure the accuracy of a simulation result, and therefore, in the measurement of the durable load of an enhanced road, the stress value and the deformation of a plate spring are required to be synchronously measured.
The deformation of the existing plate spring is often measured based on a stay wire sensor, the stress value of the plate spring is generally measured based on a force sensor, and when the posture of the whole vehicle is changed in the measuring mode, the measuring error of the deformation and the stress value is large, so that the accuracy of the subsequent judgment of the simulation result is affected.
Disclosure of Invention
The embodiment of the application provides a method for measuring the stress value and the deformation of an automobile leaf spring based on machine learning, which aims to solve the technical problems that in the prior art, the deformation and the stress value of the leaf spring are respectively measured through a stay wire sensor and a force sensor, and when the posture of the whole automobile changes, the measurement error is larger, and the accuracy of the subsequent judgment of a simulation structure is reduced.
The embodiment of the application provides a method for measuring the stress value and the deformation of an automobile leaf spring based on machine learning, which comprises the following steps:
acquiring full-load vertical load of the test plate spring in a first state, and acquiring maximum longitudinal load of the test plate spring in a second state through the full-load vertical load;
constructing a digital twin plate spring model of the test plate spring, applying the full-load vertical load to a plate spring seat of the digital twin plate spring model to obtain a maximum strain amount region on the digital twin plate spring model, applying the maximum longitudinal load to the plate spring seat of the digital twin plate spring model to obtain a minimum strain amount region on the digital twin plate spring model, and determining a position where the maximum strain amount region overlaps with the minimum strain amount region as a patch position, wherein the patch position is used for pasting a strain sheet;
bench test is carried out on the test plate spring which is pasted with the strain gauge so as to obtain a standby set formed by a plurality of standby data sets, wherein the standby data sets comprise test active load, test plate spring seat displacement and test strain values which correspond to each other;
constructing an initial neural network model, carrying out normalization processing on the standby set, and training the initial neural network model through the standby set after normalization processing to obtain a final neural network model;
and pasting the strain gauge to the patch position of the plate spring to be tested, installing the plate spring to be tested, which is pasted with the strain gauge, in the whole vehicle, and carrying out road load test on the whole vehicle to obtain an actual strain value, and determining the stress value and the deformation of the plate spring to be tested through the actual strain value and the final neural network model.
Further, the first state is a full-load static state, the second state is a braking state, and the calculation formula of the maximum longitudinal load is as follows:
，
wherein ,indicating maximum longitudinal load +.>Represents the coefficient of friction of the wheel with the ground, +.>Representing full vertical loading.
Further, the bench test is performed on the test leaf spring on which the strain gauge is posted, so as to obtain a standby set formed by a plurality of standby data sets, and the standby data sets comprise test active loads, test leaf spring seat displacement amounts and test strain values which correspond to each other, and the steps of:
placing the test plate spring with the strain gauge on a rack, applying the full-load vertical load at a plate spring seat of the test plate spring so as to enable the test plate spring to be in the first state, and resetting the measured value of the strain gauge;
applying a progressively increasing test active load at the spring seat of the test leaf spring until the test leaf spring is in a third state;
gradually reducing the test active load applied to the test leaf spring at the spring seat until the test leaf spring is in a fourth state to complete a test cycle;
and acquiring a plurality of test active loads, a plurality of test plate spring seat displacement amounts and a plurality of test strain values in the test cycle, wherein the corresponding test active loads, test plate spring seat displacement amounts and test strain values form a standby data set, and the plurality of standby data sets form the standby set.
Further, after the step of obtaining the plurality of test active loads, the plurality of test plate spring seat displacements, and the plurality of test strain values in the test cycle, the corresponding test active loads, test plate spring seat displacements, and test strain values form a standby data set, and the plurality of standby data sets form the standby set, the method further includes:
the inactive set is corrected by completing the test cycle a plurality of times.
Further, the third state is a limit compression state, and the fourth state is a free initial state.
Further, the step of constructing an initial neural network model includes:
constructing an input layer and an output layer;
constructing a first hidden layer composed of a plurality of first neurons and a second hidden layer composed of a plurality of second neurons, and connecting the first hidden layer and the second hidden layer to form a hidden layer;
and connecting the input layer with the output layer through the hidden layer to form the initial neural network model.
Further, the step of training the initial neural network model through the normalized set includes:
dividing the standby set after normalization processing into a training set, a verification set and a test set, wherein the training set, the verification set and the test set all comprise a plurality of standby data sets;
taking the test strain value in the training set as an input value, taking the test active load and the test plate spring seat displacement in the training set as output values, and training the initial neural network model once by combining a loss function to ensure the output stability of the initial neural network model;
taking the test strain value in the verification set as an input value, taking the test active load and the test plate spring seat displacement in the verification set as output values, and training the initial neural network model for the second time to judge whether the initial neural network model is over-fitted;
and training the initial neural network model for three times by taking the test strain value in the test set as an input value and taking the test active load and the test plate spring seat displacement in the test set as output values so as to determine the generalization capability of the initial neural network model.
Further, the loss function is:
，
wherein ,representing a loss function->Representing the number of inactive data sets in the training set, +.>Indicating the displacement of the test plate spring seat, +.>Indicating test active load, +.>Representing the pre-output displacement of the initial neural network model in one training, +.>Representing the pre-output load of the initial neural network model in a training.
Further, the step of determining the stress value and the deformation of the leaf spring to be measured through the actual strain value and the final neural network model specifically includes:
and inputting the actual strain value as an input value into the final neural network model to obtain an actual active load and an actual plate spring seat displacement, determining the actual active load as a stress value of the plate spring to be tested, and determining the actual plate spring seat displacement as a deformation of the plate spring to be tested.
Compared with the related art, the application has the beneficial effects that: the patch position is determined, a larger strain signal can be obtained in a subsequent bench test, a more accurate basis is provided for the back thrust of a subsequent stress value and deformation, the actual stress value can be obtained through the final neural network model when the road load test is carried out, the sensor is not required to measure, errors generated in the measuring process of the sensor are avoided, the stress value and the deformation are more accurate, and compared with the mode of the sensor, the measuring cost is greatly reduced by only using the strain gauge in the measuring process, and the automobile leaf springs of different materials can be adapted, so that the universality is improved.
The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below to provide a more thorough understanding of the other features, objects, and advantages of the application.
Drawings
FIG. 1 is a flow chart of a method for measuring stress and deformation of an automobile leaf spring based on machine learning in an embodiment of the application;
FIG. 2 is a graph showing the displacement of a test leaf spring seat obtained during bench testing in a method for measuring stress and deformation of an automobile leaf spring based on machine learning in an embodiment of the application;
FIG. 3 is a test active load obtained during bench testing in a method for measuring stress and deformation of an automobile leaf spring based on machine learning in an embodiment of the application;
FIG. 4 is a graph showing a test strain value obtained during bench testing in a method for measuring stress and deformation of an automobile leaf spring based on machine learning according to an embodiment of the present application;
FIG. 5 is a diagram of an initial neural network model in a method for measuring stress values and deformation of an automobile leaf spring based on machine learning in an embodiment of the application;
the application will be further described in the following detailed description in conjunction with the above-described figures.
Detailed Description
The present application will be described and illustrated with reference to the accompanying drawings and examples in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application. All other embodiments, which can be made by a person of ordinary skill in the art based on the embodiments provided by the present application without making any inventive effort, are intended to fall within the scope of the present application.
It is apparent that the drawings in the following description are only some examples or embodiments of the present application, and it is possible for those of ordinary skill in the art to apply the present application to other similar situations according to these drawings without inventive effort. Moreover, it should be appreciated that while such a development effort might be complex and lengthy, it would nevertheless be a routine undertaking of design, fabrication, or manufacture for those of ordinary skill having the benefit of this disclosure, and thus should not be construed as having the benefit of this disclosure.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is to be expressly and implicitly understood by those of ordinary skill in the art that the described embodiments of the application can be combined with other embodiments without conflict.
Referring to fig. 1, the method for measuring the stress value and the deformation of the automobile leaf spring based on machine learning according to the embodiment of the application comprises the following steps:
step S10: acquiring full-load vertical load of the test plate spring in a first state, and acquiring maximum longitudinal load of the test plate spring in a second state through the full-load vertical load;
the first state is a full-load static state, namely, when the test plate spring is installed on the whole vehicle, the whole vehicle is in a full-load static state; the second state is a braking state, namely, when the test plate spring is installed on the whole vehicle, the whole vehicle is in a running state. The full vertical load is the wheel load of a quarter suspension.
The calculation formula of the maximum longitudinal load is as follows:
，
wherein ,indicating maximum longitudinal load +.>Represents the coefficient of friction of the wheel with the ground, +.>Representing full vertical loading. Preferably, the friction coefficient between the wheel and the ground ranges from 0.1 to 0.2, and in this embodiment, the friction coefficient between the wheel and the ground ranges from 0.15.
Step S20: constructing a digital twin plate spring model of the test plate spring, applying the full-load vertical load to a plate spring seat of the digital twin plate spring model to obtain a maximum strain amount region on the digital twin plate spring model, applying the maximum longitudinal load to the plate spring seat of the digital twin plate spring model to obtain a minimum strain amount region on the digital twin plate spring model, and determining a position where the maximum strain amount region overlaps with the minimum strain amount region as a patch position, wherein the patch position is used for pasting a strain sheet;
the construction of the digital twin leaf spring model is performed by general industrial finite element pre-processing software or ANSA software, and the construction process is not described here in detail. The digital twin leaf spring model is highly correlated with the leaf spring. The digital twin plate spring model comprises a plate spring seat at the center, and a front coil lug and a rear coil lug which are symmetrically arranged along the plate spring seat, wherein the plate spring seat is respectively connected with the front coil lug and the rear coil lug through a rigid unit.
And the full-load vertical load is the load vertical to the direction of the plate spring seat, and is the stress value of the digital twin plate spring model at the moment for the digital twin plate spring model. When the full vertical load is applied to the leaf spring seat of the digital twin leaf spring model, strain is generated at different positions of the digital twin leaf spring model.
The maximum longitudinal load is the load in the front-rear direction of the whole vehicle, and in the digital twin plate spring model, the load in the front lug rolling direction is the load in the rear lug rolling direction. Similarly, when the maximum longitudinal load is applied to the plate spring seat of the digital twin plate spring model, strain is generated at different positions of the digital twin plate spring model.
By determining the patch position, a larger strain signal can be obtained in a subsequent bench test, and a more accurate basis is provided for the back-pushing of a subsequent stress value and deformation.
Step S30: bench test is carried out on the test plate spring which is pasted with the strain gauge so as to obtain a standby set formed by a plurality of standby data sets, wherein the standby data sets comprise test active load, test plate spring seat displacement and test strain values which correspond to each other;
and after the patch position is determined, pasting the strain gauge at the corresponding position of the test plate spring, and then performing bench test. In order to further reduce the cost when the attachment of the strain gauge is performed, the simplest strain gauge may be used.
Specifically, the step S30 includes:
s310: placing the test plate spring with the strain gauge on a rack, applying the full-load vertical load at a plate spring seat of the test plate spring so as to enable the test plate spring to be in the first state, and resetting the measured value of the strain gauge;
after the test leaf spring is placed on the bench, the state of the test leaf spring in the full-load and static state of the whole vehicle can be simulated by applying the full-load vertical load. At this time, the strain quantity measured by the strain gauge is cleared, so that the preparation work of the bench test can be completed, and the initial stage is entered.
S320: applying a progressively increasing test active load at the spring seat of the test leaf spring until the test leaf spring is in a third state;
after bench test starts, the test active load is applied to the plate spring seat of the test plate spring, so that the plate spring seat is deformed, the test active load is gradually increased, and the test plate spring reaches a limit compression state, namely, the third state is a limit compression state. In the process, along with the change of the test active load, the test plate spring seat displacement and the test strain corresponding to the test active load at a certain moment are generated on the test plate spring.
S330: gradually reducing the test active load applied to the test leaf spring at the spring seat until the test leaf spring is in a fourth state to complete a test cycle;
similarly, in this step, as the test active load changes, the test plate spring also generates a test plate spring seat displacement amount and a test strain amount corresponding to the test active load at a certain time. The fourth state is a free initial state. And after the test leaf spring is changed from the first state to the third state and then changed to the fourth state, completing one test cycle.
S340: acquiring a plurality of test active loads, a plurality of test plate spring seat displacement amounts and a plurality of test strain values in the test cycle, wherein the corresponding test active loads, test plate spring seat displacement amounts and test strain values form a standby data set, and the plurality of standby data sets form a standby set;
in this embodiment, data acquisition is performed at a sampling frequency of 512Hz in the test cycle to obtain a number of the inactive data sets. Preferably, to reduce errors in data acquisition, the step S30 further includes:
s350: the inactive set is corrected by completing the test cycle a plurality of times.
As the test cycle proceeds, the collected displacement of the test plate spring seat is shown in fig. 2, the collected test active load is shown in fig. 3, and the collected test strain value is shown in fig. 4. It will be appreciated that the test active load is the force value of the test leaf spring and the test spring seat displacement is the deflection of the test leaf spring. And providing data support for the establishment of a subsequent neural network model by acquiring the standby set.
Step S40: constructing an initial neural network model, carrying out normalization processing on the standby set, and training the initial neural network model through the standby set after normalization processing to obtain a final neural network model;
and (3) performing correlation analysis on the test strain value in the step (S30) and the test active load and the test plate spring seat displacement respectively, wherein the correlation analysis can obtain complex nonlinear correlation, if programming is performed through conventional nonlinear curve fitting, the real stress and the real deformation of the plate spring under different strains can be solved, but when the plate spring object changes each time, the stress-strain-force-displacement relation curve changes, reprogramming is needed, or the parameters of the existing codes are adjusted, so that the adjustment process is very troublesome, time-consuming and labor-consuming. The above problems can be solved by constructing the initial neural network model.
Referring to fig. 5, specifically, the step of constructing an initial neural network model includes:
s410: constructing an input layer and an output layer;
s420: constructing a first hidden layer composed of a plurality of first neurons and a second hidden layer composed of a plurality of second neurons, and connecting the first hidden layer and the second hidden layer to form a hidden layer;
s430: connecting the input layer with the output layer through the hidden layer to form the initial neural network model;
it will be appreciated that in this embodiment, the initial neural network model is constructed using a dual hidden structure, preferably, in this embodiment, the first hidden layer includes 10 first neurons and the second hidden layer includes 6 second neurons. The activation function of the hidden layer is a Relu function, and the output layer does not need to be provided with a nonlinear activation function, i.e. the data conversion from the second hidden layer to the output layer is a linear transformation. The initial neural network model selects Adam as an optimizer, and the training learning rate is selected to be 0.001.
After the initial neural network model is built, training the initial neural network model, and normalizing the standby set, namely respectively normalizing the test active load, the test plate spring seat displacement and the test strain values, so as to normalize a plurality of test active loads, a plurality of test plate spring seat displacement and a plurality of test strain values to (-1, 1), thereby improving the convergence of the initial neural network model when training the initial neural network model. Specifically, the step of training the initial neural network model through the normalized set includes:
s440: dividing the standby set after normalization processing into a training set, a verification set and a test set, wherein the training set, the verification set and the test set all comprise a plurality of standby data sets;
the training set, the verification set and the test set comprise a plurality of test active loads, test plate spring seat displacement amounts and test strain values which correspond to each other.
S450: taking the test strain value in the training set as an input value, taking the test active load and the test plate spring seat displacement in the training set as output values, and training the initial neural network model once by combining a loss function to ensure the output stability of the initial neural network model;
the loss function is:
，
wherein ,representing a loss function->Representing the number of inactive data sets in the training set, +.>Indicating the displacement of the test plate spring seat, +.>Indicating test active load, +.>Representing the pre-output displacement of the initial neural network model in one training, +.>Pre-input representing an initial neural network model in a training sessionAnd (5) discharging the load.
In the process of one training, the initial neural network model continuously adjusts the connection weight between each first neuron and each second neuron, so that the loss function is smaller and smaller, that is, the pre-output of the initial neural network model approaches to a true value, and the purpose of improving the output stability of the initial neural network model is achieved.
S460: taking the test strain value in the verification set as an input value, taking the test active load and the test plate spring seat displacement in the verification set as output values, and training the initial neural network model for the second time to judge whether the initial neural network model is over-fitted;
in this step, the initial neural network model is again trained in conjunction with the loss function. In the process of the secondary training, the purpose of judging whether the initial neural network model is over-fitted is achieved by judging whether the value of the loss function in the secondary training is higher than the value of the loss function in the primary training.
If the value of the loss function in the secondary training is higher than the value of the loss function in the primary training, the initial neural network model is over-fitted, and the step S450 needs to be performed again. Otherwise, step S470 is performed.
S470: and training the initial neural network model for three times by taking the test strain value in the test set as an input value and taking the test active load and the test plate spring seat displacement in the test set as output values so as to determine the generalization capability of the initial neural network model.
The purpose of the third training is to directly compare the pre-output and the output value of the initial neural network model, judge the generalization capability of the initial neural network model in the third training, and if the accuracy of the comparison value is poor, the number of layers of the hidden layer needs to be optimized, or the number of the first neurons and the second neurons needs to be adjusted.
Step S50: and pasting the strain gauge to the patch position of the plate spring to be tested, installing the plate spring to be tested, which is pasted with the strain gauge, in the whole vehicle, and carrying out road load test on the whole vehicle to obtain an actual strain value, and determining the stress value and the deformation of the plate spring to be tested through the actual strain value and the final neural network model.
The determination of the patch position of the leaf spring to be tested is the same as the step S20, and will not be described in detail here, it can be understood that by pasting the strain gauge, the actual strain value of the leaf spring to be tested in the road load test process of the whole vehicle can be collected, and the actual strain value is normalized.
And inputting the actual strain value as an input value into the final neural network model to obtain an actual active load and an actual plate spring seat displacement, determining the actual active load as a stress value of the plate spring to be tested, and determining the actual plate spring seat displacement as a deformation of the plate spring to be tested. And respectively reversely pushing the actual active load and the actual plate spring seat displacement to be the stress value and the deformation of the plate spring to be tested according to a normalized rule.
The final neural network model is used for testing road load, the actual strain value can be used for obtaining the stress value and the deformation, the sensor is not required to be used for measuring, errors generated in the measuring process of the sensor are avoided, the stress value and the deformation are more accurate, and compared with the mode of the sensor, the strain gauge is only required to be used in the measuring process, the measuring cost is greatly reduced, the automobile leaf springs made of different materials can be adapted, and the universality is improved.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples illustrate only a few embodiments of the application, which are described in detail and are not to be construed as limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (9)
1. The method for measuring the stress value and the deformation of the automobile leaf spring based on machine learning is characterized by comprising the following steps of:
acquiring full-load vertical load of the test plate spring in a first state, and acquiring maximum longitudinal load of the test plate spring in a second state through the full-load vertical load;
constructing a digital twin plate spring model of the test plate spring, applying the full-load vertical load to a plate spring seat of the digital twin plate spring model to obtain a maximum strain amount region on the digital twin plate spring model, applying the maximum longitudinal load to the plate spring seat of the digital twin plate spring model to obtain a minimum strain amount region on the digital twin plate spring model, and determining a position where the maximum strain amount region overlaps with the minimum strain amount region as a patch position, wherein the patch position is used for pasting a strain sheet;
bench test is carried out on the test plate spring which is pasted with the strain gauge so as to obtain a standby set formed by a plurality of standby data sets, wherein the standby data sets comprise test active load, test plate spring seat displacement and test strain values which correspond to each other;
constructing an initial neural network model, carrying out normalization processing on the standby set, and training the initial neural network model through the standby set after normalization processing to obtain a final neural network model;
and pasting the strain gauge to the patch position of the plate spring to be tested, installing the plate spring to be tested, which is pasted with the strain gauge, in the whole vehicle, and carrying out road load test on the whole vehicle to obtain an actual strain value, and determining the stress value and the deformation of the plate spring to be tested through the actual strain value and the final neural network model.
2. The method for measuring the stress value and the deformation of the automobile leaf spring based on machine learning according to claim 1, wherein the first state is a full-load static state, the second state is a braking state, and the calculation formula of the maximum longitudinal load is as follows:
，
wherein ,indicating maximum longitudinal load +.>Represents the coefficient of friction of the wheel with the ground, +.>Representing full vertical loading.
3. The method for measuring stress and deformation of a leaf spring of an automobile based on machine learning according to claim 1, wherein the step of bench testing the test leaf spring on which the strain gauge is applied to obtain a standby set of a plurality of standby data sets, the standby data sets including test active load, test leaf spring seat displacement and test strain value corresponding to each other comprises:
placing the test plate spring with the strain gauge on a rack, applying the full-load vertical load at a plate spring seat of the test plate spring so as to enable the test plate spring to be in the first state, and resetting the measured value of the strain gauge;
applying a progressively increasing test active load at the spring seat of the test leaf spring until the test leaf spring is in a third state;
gradually reducing the test active load applied to the test leaf spring at the spring seat until the test leaf spring is in a fourth state to complete a test cycle;
and acquiring a plurality of test active loads, a plurality of test plate spring seat displacement amounts and a plurality of test strain values in the test cycle, wherein the corresponding test active loads, test plate spring seat displacement amounts and test strain values form a standby data set, and the plurality of standby data sets form the standby set.
4. The method for measuring stress and deformation of a vehicle leaf spring based on machine learning according to claim 3, wherein, after the step of obtaining a plurality of the test active loads, a plurality of test leaf spring seat displacements, and a plurality of test strain values in the test cycle, the corresponding test active loads, test leaf spring seat displacements, and test strain values form a standby data set, and a plurality of the standby data sets form the standby set, further comprising:
the inactive set is corrected by completing the test cycle a plurality of times.
5. The method for measuring the stress value and the deformation amount of the automobile leaf spring based on machine learning according to claim 3, wherein the third state is a limit compression state and the fourth state is a free initial state.
6. The method for measuring stress value and deformation of an automobile leaf spring based on machine learning according to claim 1, wherein the step of constructing an initial neural network model comprises:
constructing an input layer and an output layer;
constructing a first hidden layer composed of a plurality of first neurons and a second hidden layer composed of a plurality of second neurons, and connecting the first hidden layer and the second hidden layer to form a hidden layer;
and connecting the input layer with the output layer through the hidden layer to form the initial neural network model.
7. The method for measuring stress value and deformation of automobile leaf springs based on machine learning according to claim 1, wherein the step of training the initial neural network model by the normalized set of standby comprises:
dividing the standby set after normalization processing into a training set, a verification set and a test set, wherein the training set, the verification set and the test set all comprise a plurality of standby data sets;
taking the test strain value in the training set as an input value, taking the test active load and the test plate spring seat displacement in the training set as output values, and training the initial neural network model once by combining a loss function to ensure the output stability of the initial neural network model;
taking the test strain value in the verification set as an input value, taking the test active load and the test plate spring seat displacement in the verification set as output values, and training the initial neural network model for the second time to judge whether the initial neural network model is over-fitted;
and training the initial neural network model for three times by taking the test strain value in the test set as an input value and taking the test active load and the test plate spring seat displacement in the test set as output values so as to determine the generalization capability of the initial neural network model.
8. The method for measuring the stress value and the deformation of the automobile leaf spring based on machine learning according to claim 7, wherein the loss function is:
，
wherein ,representing a loss function->Representing the number of inactive data sets in the training set, +.>Indicating the displacement of the test plate spring seat, +.>Indicating test active load, +.>Representing the pre-output displacement of the initial neural network model in one training, +.>Representing the pre-output load of the initial neural network model in a training.
9. The method for measuring the stress value and the deformation of the automobile leaf spring based on machine learning according to claim 1, wherein the step of determining the stress value and the deformation of the leaf spring to be measured by the actual strain value and the final neural network model is specifically as follows:
and inputting the actual strain value as an input value into the final neural network model to obtain an actual active load and an actual plate spring seat displacement, determining the actual active load as a stress value of the plate spring to be tested, and determining the actual plate spring seat displacement as a deformation of the plate spring to be tested.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311118723.7A CN116839783B (en) | 2023-09-01 | 2023-09-01 | Method for measuring stress value and deformation of automobile leaf spring based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311118723.7A CN116839783B (en) | 2023-09-01 | 2023-09-01 | Method for measuring stress value and deformation of automobile leaf spring based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116839783A true CN116839783A (en) | 2023-10-03 |
| CN116839783B CN116839783B (en) | 2023-12-08 |
Family
ID=88174689
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311118723.7A Active CN116839783B (en) | 2023-09-01 | 2023-09-01 | Method for measuring stress value and deformation of automobile leaf spring based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116839783B (en) |
Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FI941153A0 (en) * | 1994-03-10 | 1994-03-10 | Koivisto Marja Liisa | Method for measuring loads aimed at structures |
| KR20010010575A (en) * | 1999-07-21 | 2001-02-15 | 정몽규 | Method for estimating fatigue life of leaf spring in vehicle |
| JP2002228515A (en) * | 2001-01-30 | 2002-08-14 | Yazaki Corp | Strain sensor mounting position setting assisting method, strain sensor mounting position setting assisting device, and strain sensor mounting position setting assisting program |
| CA2484456A1 (en) * | 2004-10-12 | 2006-04-12 | Air2, Llc | Non-destructive testing of in-service wooden beams |
| CN201537687U (en) * | 2009-11-09 | 2010-08-04 | 田陆 | Self-adaptive vibrating signal acquisition device |
| CN101866378A (en) * | 2010-04-28 | 2010-10-20 | 淮阴工学院 | Method for solving rigidity of plate spring by using automatic dynamic analysis of mechanical system (ADAMS) |
| KR20120085136A (en) * | 2011-10-04 | 2012-07-31 | 주식회사 켐에쎈 | Multiple linear regression-artificial neural network model predicting ideal gas absolute entropy of pure organic compound for normal state |
| CN107480395A (en) * | 2017-08-29 | 2017-12-15 | 燕山大学 | A kind of construction method and system of vehicle steering knuckle loading spectrum forecast model |
| CN110274676A (en) * | 2019-07-22 | 2019-09-24 | 杭州鸿泉物联网技术股份有限公司 | A kind of vehicle load measurement method and device |
| CN210116420U (en) * | 2018-03-26 | 2020-02-28 | 现代自动车株式会社 | Locking robustness maintenance style sliding headrest |
| CN213543745U (en) * | 2020-10-29 | 2021-06-25 | 深圳市攀高测量技术有限公司 | Weighing sensor with prevent falling structure |
| CN113343374A (en) * | 2021-04-26 | 2021-09-03 | 江铃汽车股份有限公司 | Automobile plate spring fatigue testing method |
| WO2022083009A1 (en) * | 2020-10-20 | 2022-04-28 | 浙江大学 | Customized product performance prediction method based on heterogeneous data error compensation fusion |
| CN114595721A (en) * | 2022-03-09 | 2022-06-07 | 上海理工大学 | Vehicle part strain load prediction method based on road condition recognition |
| US20220196459A1 (en) * | 2020-12-22 | 2022-06-23 | Dibi (Chongqing) Intelligent Technology Research Institute Co., Ltd. | Real-time vehicle overload detection method based on convolutional neural network |
| CN114879536A (en) * | 2022-07-11 | 2022-08-09 | 山东交通学院 | Method and device for acquiring real-time characteristics of suspension system based on digital twinning technology |
| CN115169167A (en) * | 2022-05-26 | 2022-10-11 | 江铃汽车股份有限公司 | Method and system for optimizing and matching motion stroke parameters of automobile plate spring |
| CN115270284A (en) * | 2022-05-23 | 2022-11-01 | 中国第一汽车股份有限公司 | Shock absorber dynamic characteristic identification model establishing method and simulation method |
| WO2023045339A1 (en) * | 2021-09-27 | 2023-03-30 | 大连理工大学 | Weld fatigue digital twin framework based on structural stress method |
| CN115935554A (en) * | 2022-12-16 | 2023-04-07 | 中国重汽集团济南动力有限公司 | Design method and design terminal for multi-level-rigidity rear plate spring of light truck |
| CN116090111A (en) * | 2023-04-10 | 2023-05-09 | 华东交通大学 | Automobile leaf spring fatigue life prediction method based on deep learning model |
| WO2023115596A1 (en) * | 2021-12-21 | 2023-06-29 | 浙江工业大学台州研究院 | Truss stress prediction and weight lightening method based on transfer learning fusion model |
| CN116579209A (en) * | 2023-05-11 | 2023-08-11 | 南京航空航天大学 | Wing structure deformation real-time prediction method based on digital twin and neural network |
-
2023
- 2023-09-01 CN CN202311118723.7A patent/CN116839783B/en active Active
Patent Citations (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FI941153A0 (en) * | 1994-03-10 | 1994-03-10 | Koivisto Marja Liisa | Method for measuring loads aimed at structures |
| KR20010010575A (en) * | 1999-07-21 | 2001-02-15 | 정몽규 | Method for estimating fatigue life of leaf spring in vehicle |
| JP2002228515A (en) * | 2001-01-30 | 2002-08-14 | Yazaki Corp | Strain sensor mounting position setting assisting method, strain sensor mounting position setting assisting device, and strain sensor mounting position setting assisting program |
| CA2484456A1 (en) * | 2004-10-12 | 2006-04-12 | Air2, Llc | Non-destructive testing of in-service wooden beams |
| CN201537687U (en) * | 2009-11-09 | 2010-08-04 | 田陆 | Self-adaptive vibrating signal acquisition device |
| CN101866378A (en) * | 2010-04-28 | 2010-10-20 | 淮阴工学院 | Method for solving rigidity of plate spring by using automatic dynamic analysis of mechanical system (ADAMS) |
| KR20120085136A (en) * | 2011-10-04 | 2012-07-31 | 주식회사 켐에쎈 | Multiple linear regression-artificial neural network model predicting ideal gas absolute entropy of pure organic compound for normal state |
| CN107480395A (en) * | 2017-08-29 | 2017-12-15 | 燕山大学 | A kind of construction method and system of vehicle steering knuckle loading spectrum forecast model |
| CN210116420U (en) * | 2018-03-26 | 2020-02-28 | 现代自动车株式会社 | Locking robustness maintenance style sliding headrest |
| CN110274676A (en) * | 2019-07-22 | 2019-09-24 | 杭州鸿泉物联网技术股份有限公司 | A kind of vehicle load measurement method and device |
| WO2022083009A1 (en) * | 2020-10-20 | 2022-04-28 | 浙江大学 | Customized product performance prediction method based on heterogeneous data error compensation fusion |
| CN213543745U (en) * | 2020-10-29 | 2021-06-25 | 深圳市攀高测量技术有限公司 | Weighing sensor with prevent falling structure |
| US20220196459A1 (en) * | 2020-12-22 | 2022-06-23 | Dibi (Chongqing) Intelligent Technology Research Institute Co., Ltd. | Real-time vehicle overload detection method based on convolutional neural network |
| CN113343374A (en) * | 2021-04-26 | 2021-09-03 | 江铃汽车股份有限公司 | Automobile plate spring fatigue testing method |
| WO2023045339A1 (en) * | 2021-09-27 | 2023-03-30 | 大连理工大学 | Weld fatigue digital twin framework based on structural stress method |
| WO2023115596A1 (en) * | 2021-12-21 | 2023-06-29 | 浙江工业大学台州研究院 | Truss stress prediction and weight lightening method based on transfer learning fusion model |
| CN114595721A (en) * | 2022-03-09 | 2022-06-07 | 上海理工大学 | Vehicle part strain load prediction method based on road condition recognition |
| CN115270284A (en) * | 2022-05-23 | 2022-11-01 | 中国第一汽车股份有限公司 | Shock absorber dynamic characteristic identification model establishing method and simulation method |
| CN115169167A (en) * | 2022-05-26 | 2022-10-11 | 江铃汽车股份有限公司 | Method and system for optimizing and matching motion stroke parameters of automobile plate spring |
| CN114879536A (en) * | 2022-07-11 | 2022-08-09 | 山东交通学院 | Method and device for acquiring real-time characteristics of suspension system based on digital twinning technology |
| CN115935554A (en) * | 2022-12-16 | 2023-04-07 | 中国重汽集团济南动力有限公司 | Design method and design terminal for multi-level-rigidity rear plate spring of light truck |
| CN116090111A (en) * | 2023-04-10 | 2023-05-09 | 华东交通大学 | Automobile leaf spring fatigue life prediction method based on deep learning model |
| CN116579209A (en) * | 2023-05-11 | 2023-08-11 | 南京航空航天大学 | Wing structure deformation real-time prediction method based on digital twin and neural network |
Non-Patent Citations (6)
| Title |
|---|
| DIXIT, N.等: "Enhanced SAE 3 link Leaf Spring Model to Generate Durability Virtual Loads", SAE TECHNICAL PAPERS * |
| MOON, ID等: "A flexible multi-body dynamic model for analyzing the hysteretic characteristics and the dynamic stress of a taper leaf spring", JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY, vol. 20, no. 10 * |
| 张秋峰;刘夫云;杨超;王秋花;赵亮亮;吴鹏兴;: "载货汽车的板簧刚度验证与平顺性仿真分析", 噪声与振动控制, no. 01 * |
| 贾慧利;: "基于道路载荷谱的汽车后减振器座失效研究", 西安文理学院学报(自然科学版), no. 02 * |
| 钱琛: "某轻型客车复合材料板簧关键特性建模与性能优化", 中国博士学位论文全文数据库（电子期刊）工程科技II辑, no. 1 * |
| 韩为铎;卢剑伟;莫家奇;: "非线性渐变刚度钢板弹簧疲劳分析载荷谱编制方法", 振动与冲击, no. 06 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116839783B (en) | 2023-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108133082B (en) | Method for determining stress measurement constant in indentation strain method based on finite element simulation | |
| CN116090111B (en) | Automobile leaf spring fatigue life prediction method based on deep learning model | |
| CN111735645B (en) | Load compilation method for durability test of automobile stabilizer bar rack | |
| CN111638034A (en) | Strain balance temperature gradient error compensation method and system based on deep learning | |
| CN116839783B (en) | Method for measuring stress value and deformation of automobile leaf spring based on machine learning | |
| CN116412995A (en) | Method for determining wind tunnel strain balance measurement uncertainty among partitions | |
| US20210300391A1 (en) | System and method for measuring road surface input load for vehicle | |
| CN111191391B (en) | Simulation piece design method based on local stress distribution weighting coefficient | |
| CN113221351A (en) | CAE analysis method for surface rigidity of passenger car door | |
| CN111380649A (en) | Vehicle multi-dimensional force sensor decoupling method based on PSO (particle swarm optimization) optimization LSSVM (least squares support vector machine) | |
| CN115356104A (en) | Parameterized evaluation and prediction method for hub bearing rigidity | |
| CN114993604B (en) | Wind tunnel balance static calibration and measurement method based on deep learning | |
| CN115526276A (en) | Wind tunnel balance calibration load prediction method with robustness | |
| CN111090959A (en) | Vehicle load spectrum acquisition method and system | |
| CN114254533B (en) | Method for examining influence and prediction of fatigue vibration on fixed angle of product group component | |
| CN114970237A (en) | Method for improving fatigue endurance simulation precision of stabilizer bar system | |
| CN112179550B (en) | Optimized arrangement method of four-fulcrum piezoelectric dynamometer | |
| CN115048726A (en) | Method and system for calculating maximum stress distribution of vehicle chassis parts and storage medium | |
| CN110987269B (en) | Method and system for determining position of strain gauge of force measuring wheel set | |
| CN111623909B (en) | Stress test method for double-fork-arm two-force rod | |
| CN112733399A (en) | Failure evaluation method and device for bolt in random vibration | |
| CN116842663B (en) | Machine learning-based nonlinear time-varying stress value measuring method for automobile leaf springs | |
| CN110501177A (en) | Cantilever beam damnification recognition method based on free end inclination effect line curvature | |
| CN114647897B (en) | Cab structure optimization method based on hierarchical entropy-gray correlation method | |
| CN112632834B (en) | Failure evaluation method and device for bolt in earthquake |
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 |