CN111238927A - Fatigue durability evaluation method and device, electronic equipment and computer readable medium - Google Patents
Fatigue durability evaluation method and device, electronic equipment and computer readable medium Download PDFInfo
- Publication number
- CN111238927A CN111238927A CN201911424498.3A CN201911424498A CN111238927A CN 111238927 A CN111238927 A CN 111238927A CN 201911424498 A CN201911424498 A CN 201911424498A CN 111238927 A CN111238927 A CN 111238927A
- Authority
- CN
- China
- Prior art keywords
- fatigue
- characteristic information
- stress
- model
- fatigue failure
- 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
- 238000011156 evaluation Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 claims abstract description 103
- 230000008569 process Effects 0.000 claims abstract description 63
- 238000012545 processing Methods 0.000 claims abstract description 38
- 238000003062 neural network model Methods 0.000 claims abstract description 23
- 230000009347 mechanical transmission Effects 0.000 claims abstract description 16
- 230000033001 locomotion Effects 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 61
- 238000012549 training Methods 0.000 claims description 26
- 230000009466 transformation Effects 0.000 claims description 24
- 238000000605 extraction Methods 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 230000008878 coupling Effects 0.000 claims description 11
- 238000010168 coupling process Methods 0.000 claims description 11
- 238000005859 coupling reaction Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 abstract description 15
- 238000001514 detection method Methods 0.000 abstract description 12
- 230000015556 catabolic process Effects 0.000 abstract description 7
- 238000006731 degradation reaction Methods 0.000 abstract description 7
- 238000009661 fatigue test Methods 0.000 abstract description 4
- 238000004806 packaging method and process Methods 0.000 abstract description 3
- 230000035882 stress Effects 0.000 description 40
- 238000012360 testing method Methods 0.000 description 28
- 238000005516 engineering process Methods 0.000 description 9
- 230000004913 activation Effects 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 125000004122 cyclic group Chemical group 0.000 description 4
- 238000010998 test method Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000006073 displacement reaction Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000001931 thermography Methods 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000003638 chemical reducing agent Substances 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000012886 linear function Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004904 shortening Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 241000039077 Copula Species 0.000 description 1
- 229910000677 High-carbon steel Inorganic materials 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002929 anti-fatigue Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- FFBHFFJDDLITSX-UHFFFAOYSA-N benzyl N-[2-hydroxy-4-(3-oxomorpholin-4-yl)phenyl]carbamate Chemical compound OC1=C(NC(=O)OCC2=CC=CC=C2)C=CC(=C1)N1CCOCC1=O FFBHFFJDDLITSX-UHFFFAOYSA-N 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 229910003460 diamond Inorganic materials 0.000 description 1
- 239000010432 diamond Substances 0.000 description 1
- 238000000695 excitation spectrum Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009916 joint effect Effects 0.000 description 1
- 238000003754 machining Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000005482 strain hardening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N3/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N3/02—Details
- G01N3/06—Special adaptations of indicating or recording means
- G01N3/068—Special adaptations of indicating or recording means with optical indicating or recording means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/41—Refractivity; Phase-affecting properties, e.g. optical path length
- G01N21/45—Refractivity; Phase-affecting properties, e.g. optical path length using interferometric methods; using Schlieren methods
-
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/0058—Kind of property studied
- G01N2203/0069—Fatigue, creep, strain-stress relations or elastic constants
- G01N2203/0073—Fatigue
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/0058—Kind of property studied
- G01N2203/0069—Fatigue, creep, strain-stress relations or elastic constants
- G01N2203/0075—Strain-stress relations or elastic constants
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
- G01N2203/02—Details not specific for a particular testing method
- G01N2203/06—Indicating or recording means; Sensing means
- G01N2203/0641—Indicating or recording means; Sensing means using optical, X-ray, ultraviolet, infrared or similar detectors
- G01N2203/0647—Image analysis
Abstract
The invention provides a fatigue durability evaluation method, a device, electronic equipment and a computer readable medium, which relate to the technical field of fatigue tests of mechanical transmission members, and comprise the steps of obtaining a stress-strain electronic speckle image of a tested object, and reading fatigue process characteristic information according to the stress-strain electronic speckle image; therefore, a pre-trained target neural network model is adopted for analyzing and processing the stress-strain electronic speckle image; obtaining a data prediction result of the tested component, and predicting the durability of the tested component according to the data prediction result; the invention establishes an effective fatigue failure prediction method, improves the accuracy and the real-time performance of non-contact detection, expands the fatigue durability evaluation database and the expert system, can be suitable for transmission component occasions with the characteristics of relative motion and internal packaging, comprehensively reflects the factors of load history, strength degradation, structural characteristics and the like, and better characterizes and describes uncertain factors existing in engineering practice.
Description
Technical Field
The invention relates to the technical field of mechanical transmission member fatigue testing, in particular to a method and a device for evaluating fatigue durability, electronic equipment and a computer readable medium.
Background
The transmission component is often the main stressed component in the mechanical system, the fatigue failure has universality, instantaneity and importance, and due to the characteristics of relative movement, internal packaging and the like, the complexity and the difficulty of detecting and evaluating the fatigue durability of the transmission component are increased, so that the optimization direction of the fatigue-resistant design is difficult to define.
At present, in the fatigue durability evaluation of mechanical transmission components, a fatigue reliability model of the mechanical transmission components usually lacks consideration of the correlation of failure modes, multi-part damage and the correlation among multiple components, fails to comprehensively reflect the factors such as load history, strength degradation and structural features, and cannot well represent and describe uncertain factors existing in engineering practice.
Disclosure of Invention
The invention aims to provide a fatigue durability evaluation method, a fatigue durability evaluation device, electronic equipment and a computer readable medium, which can comprehensively reflect the factors such as load history, strength degradation, structural characteristics and the like, and better characterize and describe uncertain factors existing in engineering practice.
In a first aspect, the present invention provides a fatigue durability testing method applied to a mechanical transmission member, including:
acquiring fatigue failure characteristic information and fatigue process characteristic information of a measured component, wherein the fatigue failure characteristic information is used for representing the fatigue limit state of the measured component, and the fatigue process characteristic information is used for representing the motion process state of the measured component before the measured component is in the fatigue limit state and fails;
acquiring a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
analyzing and processing the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
In an optional embodiment, the obtaining fatigue failure characteristic information of the measured component comprises:
constructing a multi-influence parameter fatigue failure relation model of the measured component;
and determining characteristic information of the fatigue limit state of the tested component according to the multi-influence parameter fatigue failure relation model. In an optional embodiment, the multiple influence parameter fatigue failure relation model includes a fatigue failure function linear approximation model and a fatigue failure function nonlinear approximation model, and constructing the multiple influence parameter fatigue failure relation model of the measured component includes:
determining the quantitative relation of a single influence parameter of the measured object and the coupling correlation nonlinear relation of a plurality of influence parameters;
constructing a fatigue failure function linear approximation model of the measured component according to the quantitative relation of the single influence parameter;
and constructing a fatigue failure function nonlinear model of the tested member according to the coupling association nonlinear relation of the plurality of influence parameters.
In an optional embodiment, the obtaining of the fatigue process characteristic information of the measured component comprises:
constructing a fatigue durability universal model of the tested component;
and determining the fatigue process characteristic information of the tested component according to the fatigue durability universal model.
In an alternative embodiment, a stress-strain electronic speckle image of the measured member is acquired; obtaining failure characteristic information and fatigue process characteristic information according to the stress-strain electronic speckle image comprises the following steps:
recognizing and reading the stress-strain electronic speckle image by using a machine vision fuzzy algorithm based on a neural network to obtain stress and/or strain amplitude;
carrying out image distortion calibration processing on the stress strain electronic speckle image, and correcting image distortion caused by the lens precision or the assembly process of the acquisition system to obtain a calibrated stress strain electronic speckle image;
performing color space transformation on HSV and HLS in the calibrated stress-strain electronic speckle image, and obtaining a graying processing result by adopting a weighted average method; performing edge extraction on the gray processing result by using a Canny operator to obtain an edge extraction result;
selecting a target channel in a color space based on the graying processing result and the edge extraction result; determining the target channel as a binaryzation reference channel; performing binarization processing and perspective transformation on the stress strain electronic speckle image based on the binarized reference channel to obtain a transformation result;
and performing parameter fitting based on the transformation result to obtain reconstruction characteristic information of the stress-strain electronic speckle image, and determining the reconstruction characteristic information as the fatigue process characteristic information.
In an alternative embodiment, the method further comprises:
obtaining a training sample;
training the initial stage of the target neural network model based on the training samples, and deriving and updating the weight and the bias term of the convolution kernel through a loss function to obtain the global optimal weight of the target neural network model.
In an alternative embodiment, the method further comprises:
after the initial stage of the target neural network model is trained based on the training samples, the recall ratio and/or precision ratio of the training results are evaluated, and the optimal learning rate required by the model is obtained from the two.
In a second aspect, the present invention provides a fatigue durability evaluation device for a mechanical transmission member, including:
the device comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring fatigue failure characteristic information and fatigue process characteristic information of a measured component, the fatigue failure characteristic information is used for representing a fatigue limit state of the measured component, and the fatigue process characteristic information is used for representing a motion process state of the measured component before the measured component fails in the fatigue limit state; the second acquisition module is used for acquiring a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
the processing module is used for analyzing and processing the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
In a third aspect, an embodiment provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the steps of the method of any one of the foregoing embodiments when executing the computer program.
In a fourth aspect, embodiments provide a computer readable medium having non-volatile program code executable by a processor, the program code causing the processor to perform the method of any of the preceding embodiments.
According to the fatigue durability evaluation method, the fatigue durability evaluation device, the electronic equipment and the computer readable medium, the fatigue process characteristic information is obtained by obtaining the stress-strain electronic speckle image of the tested component; analyzing and processing the fatigue process characteristic information according to a pre-trained target neural network model to obtain a data prediction result of the measured component, and predicting the durability of the measured component according to the data prediction result; the invention can comprehensively reflect the factors of load course, strength degradation, structural characteristics and the like, and better characterize and describe uncertain factors existing in engineering practice.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method for fatigue endurance evaluation according to an embodiment of the present invention;
FIG. 2 is another flow chart of a method for fatigue endurance evaluation according to an embodiment of the present invention;
FIG. 3 is a speckle pattern before and after deformation of an object in the same coordinate system according to the fatigue durability evaluation method provided by the embodiment of the invention;
FIG. 4 is a schematic diagram illustrating comparison before and after calibration of a distorted image according to the fatigue durability evaluation method provided by the embodiment of the invention;
FIG. 5 is a schematic diagram of a fatigue endurance testing apparatus according to an embodiment of the present invention;
fig. 6 is a system schematic diagram of an electronic device according to an embodiment of the present invention.
Icon: 31-a first acquisition module; 32-a second acquisition module; 33-a processing module; 400-an electronic device; 401 — a communication interface; 402-a processor; 403-a memory; 404-bus.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the following embodiments, and it should be understood that the described embodiments are some, but not all, embodiments of the present invention. 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.
The fatigue durability evaluation technical characteristics of the current mechanical transmission component are generally expressed as follows: the analysis method is complex, the boundary condition is difficult to be clear, the influence factors are cross-coupled, the fault tree system of applicability is imperfect, the optimization direction of the anti-fatigue design is dispersed, and the like; the test has the advantages of huge distribution of parameter measuring points, numerous types of subsystems, long time period, large resource consumption, high manual participation degree and difficult test means; and the evaluation method is not standard and the autonomous task mode is not uniform.
The transmission component is often the main stressed component in the mechanical system, the fatigue failure has universality, instantaneity and importance, and due to the characteristics of relative movement, internal packaging and the like, the complexity and the difficulty of detecting and evaluating the fatigue durability of the transmission component are increased, so that the optimization direction of the fatigue-resistant design is difficult to define. Currently, the existing fatigue durability test has the following disadvantages:
the fatigue reliability model of the mechanical transmission component usually lacks the relevance considering the failure mode, multi-part damage and the correlation among multiple components, fails to comprehensively reflect the factors such as load course, strength degradation, structural features and the like, and cannot well represent and describe the uncertain factors existing in engineering practice.
The fatigue durability technical standard of the mechanical transmission component is not standard, the testing and failure prediction means are single, the manual on-duty detection method is still relied on, the full applicability research and testing verification are lacked, and the service life distribution and failure prediction method cannot be effectively established; the fatigue detection level of mechanical components is low, the matching performance of instruments is poor, the reliability of detection results is low, comparability is lacked among the instruments, test items cannot meet the requirement of product quality control, and the development of the mechanical equipment industry is severely restricted.
The accuracy and the real-time performance of a non-contact detection technology and a deep learning technology based on machine vision need to be improved, the application expansion of the non-contact detection technology and the deep learning technology in the fatigue characteristic detection and estimation of mechanical transmission members is limited, and a function integration and fault prediction model of the non-contact detection technology and the deep learning technology is not used for targeted theory reference.
A fatigue durability evaluation database or expert system lacking versatility and accuracy.
Based on the method, the device, the electronic equipment and the computer readable medium, the fatigue durability test method, the device, the electronic equipment and the computer readable medium can comprehensively reflect elements such as load process, strength degradation, structural characteristics and the like, and can well represent and describe uncertain factors existing in engineering practice. The details are described below by way of examples.
Referring to fig. 1, the fatigue durability test method provided by the embodiment of the invention is applied to a mechanical transmission component, and comprises the following steps:
s110, acquiring fatigue failure characteristic information and fatigue process characteristic information of a measured component, wherein the fatigue failure characteristic information is used for representing the fatigue limit state of the measured component, and the fatigue process characteristic information is used for representing the motion process state of the measured component before the fatigue limit state fails;
s120, acquiring a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
s130, analyzing and processing the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
Specifically, as shown in fig. 2, with the image capturing system of the present embodiment, for example: the CCD camera, the infrared thermal imaging camera, the laser camera electronic speckle pattern interference technology (ESPI) and the digital image technology (DIC) carry out data acquisition on a mechanical transmission member (namely the measured object in the embodiment), and acquire a reference image before a test, a micro deformation characteristic image before fatigue failure and a fatigue failure characteristic image.
In systems such as automobile gearboxes, transmission shafts or robot reducers, mechanical arms and the like, transmission components are subjected to periodic or aperiodic alternating load for a long time, the accumulation of internal stress and strain easily causes the initiation, the expansion and even the fracture of cracks, and the damage caused by the fatigue damage is called fatigue failure. The mechanical components can still reach the limit of corresponding technical or economic indexes under the influence of fatigue factors, and the capability of completing the preset transmission function is called fatigue durability.
The stress strain electronic speckle images of systems such as an automobile gearbox, a transmission shaft or a robot speed reducer, a mechanical arm and the like are obtained, so that fatigue failure characteristic information and fatigue durability information can be obtained.
The target neural network model of the embodiment is a convolutional neural network model, the early failure image is identified through an SSD fuzzy algorithm, and the fatigue process characteristic information is analyzed and processed through the convolutional neural network model to obtain a data prediction result of the measured component.
The convolutional neural network is a feedforward neural network which comprises convolutional calculation and has a deep structure, is one of main algorithms of deep learning, and mainly comprises a convolutional layer, an activation layer, a pooling layer, a residual layer, a full-link layer or a route layer and the like. By introducing the concept of convolution, weight sharing can be realized, the size and the dimensionality of an output image are controlled by controlling the sliding step length of a convolution kernel and the number of the convolution kernels, and the number of parameters of a network is further reduced. The defect of the weight of the fully-connected neural network is greatly improved, and meanwhile, the occurrence of overfitting of the fully-connected neural network is effectively avoided. Two-dimensional convolutional neural networks are commonly used for processing images, but odd determinant convolutional kernels of 1 × 1, 3 × 3, 5 × 5 and the like are selected for realization.
The construction process of the target neural network model is as follows:
(1) determining activation functions
In the transmission process of the neural network, the input is not linearly separable under most conditions, so the activation function (activation function) is used for introducing a non-linear function which often follows a convolutional layer. Common activation functions are Sigmoid function, Softmax function, tanh function, ReLU function, etc.
softmax function:eifor the ith output value of the neural network, sigmajejIs the output value of the j-dimensional neural network.
ReLU function: f. ofR(x)=max(0,x);
(2) Non-maximum suppression
In the transmission process of the neural network, the input is not linearly separable under most conditions, so the activation function (activation function) is used for introducing a non-linear function which often follows a convolutional layer. Common activation functions are Sigmoid function, Softmax function, tanh function, ReLU function, etc.
softmax function:eifor the ith output value of the neural network, sigmajejIs the output value of the j-dimensional neural network.
ReLU function: f. ofR(x)=max(0,x);
Network training is carried out, for the output of the model, a plurality of preselected frames may exist for the detection of one object, and then a Non Maximum Suppression (NMS) method is required to be used for processing the preselected frames. For the mutually overlapped preselected frames, the preselected frame with the largest prediction probability is selected as a reference frame, and the ratio of the area calculated by combining the preselected frames intersected with the reference frame to the area calculated by combining the preselected frames, namely the combining ratio, is calculated. If the obtained value is larger than a certain threshold value, the overlapping degree of the preselected frame and the reference frame is considered to be high and deleted, otherwise, the following steps are retained:
Sp1、Sp2the areas of the two prediction boxes are respectively represented.
(3) Data outcome prediction
In order to find the balance point, different threshold values need to be set, the precision ratio and the recall ratio corresponding to each threshold value are found, and all the points are drawn into the PR curve of the model. The better the model, the larger the area enclosed by the PR curve with the x-axis and the y-axis. This area was taken as the Average accuracy of the model (AP):
the formula can predict the AP value of a certain image, and for the problem of multi-classification and multi-information, the AP average value of multiple images needs to be obtained.
The embodiment can comprehensively reflect the factors such as load history, strength degradation, structural characteristics and the like through the fatigue process characteristic information, and well characterize and describe uncertain factors existing in engineering practice.
Optionally, the obtaining of the fatigue failure characteristic information of the tested member in the fatigue durability testing method according to the embodiment includes:
constructing a multi-influence parameter fatigue failure relation model of the measured component;
and determining characteristic information of the fatigue limit state of the tested component according to the multi-influence parameter fatigue failure relation model.
Specifically, a multi-influence-parameter fatigue failure relation model of a measured component is established by a single parameter and multiple parameters, since fatigue is a process, the fatigue process is influenced by multiple parameters, and some parameters have coupling properties, for convenience of evaluation, analysis is usually performed from the single parameter first, and other parameters are set under specific conditions (that is, it is assumed that under the conditions, changes among the parameters do not influence changes of an overall index). Then, the coupling and the relevance among the multiple parameters (namely the effect of the joint action of the multiple parameters) are considered, but the coupling makes the analysis difficult. Therefore, the embodiment needs to acquire a multiple-influence-parameter fatigue failure relationship model to determine the fatigue failure characteristic information, so that the whole fatigue durability test (long cycle, complicated coupling of influence parameters) of the component is detected and evaluated (i.e. evaluated) through steps S110 to S120 of the embodiment. All fatigue characteristics can be reflected on the stress-strain electronic speckle images, so that the fatigue failure information and the fatigue durability information need to be obtained through the stress-strain electronic speckle images.
Optionally, the multiple-influence-parameter fatigue failure relationship model includes a fatigue failure function linear approximation model and a fatigue failure function nonlinear approximation model, and the constructing the multiple-influence-parameter fatigue failure relationship model of the measured component in the above embodiment includes:
determining the quantitative relation of a single influence parameter of the measured object and the coupling correlation nonlinear relation of a plurality of influence parameters;
constructing a fatigue failure function linear approximation model of the measured component according to the quantitative relation of the single influence parameter;
and constructing a fatigue failure function nonlinear model of the tested member according to the coupling association nonlinear relation of the plurality of influence parameters.
Specifically, the single influencing parameter includes at least: material properties, dimensional structure, surface roughness, stress concentration, heat treatment and the like. Quantitative analysis of the single influencing parameters was carried out by Miner's rule. Specifically, the analysis was performed according to the following formula:
wherein, it is assumed that a single parameter corresponds to the stress SiAre independent of each other. n isiThe number of cycles is accumulated for fatigue. N is a radical ofiTo be under stress SiFatigue limit life. SfThe fatigue strength is 0.69 of the low-carbon diamond; taking 0.68 of common steel; the high carbon steel is 0.67. F () is the fatigue strength function, S is the loading stress, KtFor theoretical stress concentration coefficient,. epsilon.for smooth specimen size coefficient, β1The surface machining coefficient of the part is shown.
The linear accumulated fatigue failure refers to that under the action of cyclic stress, fatigue and the number of load cycles are in a linear relation, fatigue damage among all levels of stress is independent and does not influence each other, and the sum of the fatigue failures of loads with different sizes is equal to the linear accumulation of the action of each load. When a certain limit is breached, the component will fail. The linear approximation model of the fatigue failure function of the constructed tested member is shown as the following formula:
for a plurality of influencing parameters, the fatigue limit of a transmission component is influenced by a plurality of external uncertain factors, and fatigue failure is inevitably caused to have certain fuzzy characteristics in the whole working process due to different parameters such as stress condition, processing quality, appearance structure size, surface quality, load loading sequence and the like.
Introducing a general expression of the fuzzy Miner rule, and considering a probability S-N curve to obtain the fuzzy probability Miner rule:
wherein Si(i ═ 1,2,3, …, k) are subject to different stress levels of class k. The k-m level is below the fatigue limit. N is a radical ofpiThe fatigue failure life with the reliability p under the independent action of the ith-level stress is shown.The fatigue limit life at the reliability probability p.Is a membership function.
Wherein high load-low load:
low load-high load time:
wherein SRThe fatigue limit. (S)R)H-LRepresents the fuzzy lower bound of fatigue limit under high-low loading. (S)R)L-HRepresents the fuzzy upper bound of fatigue limit under low-high loading.
The mutual influence among the loads has remarkable nonlinear characteristics, and the function causing the fatigue failure also has nonlinearity. For engineering application, the introduced analysis model considers the influence of the sequence among loads on the fatigue strength and the influence of the interaction on the fatigue life prediction, and the calculation parameters are simplified, so that the fatigue failure function nonlinear model of the measured component is constructed by the following formula:
whereinFor each stage of cyclic ratio, niNumber of stress cycles, NfIs the fatigue life of the applied load. N is a radical offiThe fatigue life at each level of loading.
Optionally, the fatigue durability testing method of the above embodiment further includes:
constructing a fatigue durability universal model of the tested component;
and determining the characteristic information of the fatigue gradual change process of the tested component according to the fatigue durability universal model.
Specifically, a fatigue durability general model is constructed by a nominal stress method (S-N method) and a local stress strain method (e-N method).
S-N method:
e-N method:
wherein epsilonaIs the strain magnitude.Is the elastic strain amplitude.Is the magnitude of the plastic strain. SigmaaIs the stress magnitude. E is the modulus of elasticity. K' is the cyclic intensity coefficient. n' is the cyclic strain hardening index. Sigma'fThe fatigue strength coefficient. b is fatigue strength index. Epsilon'fThe fatigue ductility factor. c is fatigue ductility index.
Optionally, step S110 in the above embodiment includes:
recognizing and reading the stress-strain electronic speckle image by using a machine vision fuzzy algorithm based on a neural network to obtain stress and/or strain amplitude;
carrying out image distortion calibration processing on the stress strain electronic speckle image, and correcting image distortion caused by the lens precision or the assembly process of an acquisition system;
performing color space transformation on HSV and HLS in the stress strain electronic speckle image, and obtaining a graying processing result by adopting a weighted average method; performing edge extraction on the gray processing result by using a Canny operator to obtain an edge extraction result;
selecting a target channel in a color space based on the graying processing result and the edge extraction result; determining the target channel as a binaryzation reference channel; performing binarization processing and perspective transformation on the stress strain electronic speckle image based on the binarized reference channel to obtain a transformation result;
and performing parameter fitting based on the transformation result to obtain reconstruction characteristic information of the stress-strain electronic speckle image, and determining the reconstruction characteristic information as the fatigue process characteristic information.
Specifically, the machine vision fuzzy algorithm based on the neural network mainly comprises: the method comprises the steps of digital speckle measurement correlation algorithm, image distortion calibration, color space transformation and edge extraction, image binarization and perspective transformation and parameter fitting.
Wherein, referring to fig. 3, a small area speckle distribution around each point on the speckle pattern is different from other points, thereby forming a subset. The subset with a certain point as the center can be used as an information carrier of the displacement of the point, and the corresponding information such as deformation, stress or strain can be obtained by analyzing and searching the movement and the change of the subset.
Object composed of1(x, y) deformation to S2(x ', y'), taking a certain pixel (m × m) area A before deformation as a matching template, and performing correlation operation on the corresponding pixel (m × m) area B after deformation:
c is the correlation coefficient of the two subsets. f. ofA(xi,yj)、fB(x′i,y′j) As a function of the gray set of regions a and B.Is the combined mean of A and B, i.e.
Referring to fig. 4, the principle of performing image distortion calibration processing on the stress-strain electronic speckle image is as follows:
the imaging process of the camera is actually to transform points in space from the "world coordinate system" to the "camera coordinate system" and then project them to the "physical coordinate system" of the image. However, due to the problems of the precision and the assembly process of the lens, the shot image is distorted, and the image is distorted. For this purpose, the image needs to be calibrated before it is recognized. The distortion may be divided into radial distortion and tangential distortion according to the cause of the distortion.
Radial distortion: the rays are more curved at the center of the lens than around the center.
Taking the coordinates of the center point (x)c,yc) The peripheral Taylor series expansion:
where (x, y) is the coordinate of a certain point of the distorted original image. (x)c,yc) The center point coordinates after distortion calibration. (x)ca,yca) For the (x, y) corresponding point coordinates after calibration,k1、k2for the lens radial distortion coefficient, only the corresponding (k) needs to be found in the calibration process1,k2) And (4) finishing.
Tangential distortion: due to the fact that the lens and the sensor plane of the camera are not parallel.
Wherein p is1、p2Is the lens tangential distortion coefficient.
The principle of color space transformation is:
graying of HSV (hue, saturation, hue value) and HLS (hue, brightness, saturation) color spaces in electronic speckle images can be generally converted by RGB (Red, Green, Blue) channels.
Let max (R, G, B) be the maximum of R, G, B and min (R, G, B) be the minimum.
RGB-HSV:
V=max(R,G,B);
RGB-HLS, hue H is the same:
obtaining a gray scale image by adopting a weighted average method:
f(i,j)=0.3R(i,j)+0.59G(i,j)+0.11B(i,j);
wherein, R (i, j) is the value of the channel R at the coordinate (i, j), and G (i, j) and B (i, j) are the same as above.
The principle of edge extraction is as follows:
and (4) selecting a Canny operator to carry out edge extraction, which is a dual-threshold edge detection algorithm. First, a two-dimensional convolution kernel is generated using the gaussian formula:
and then carrying out convolution filtering on the image subjected to the previous graying:
fS(x,y)=f(x,y)+G(x,y);
gradient values are calculated for the convolved image using first order differences:
where dx [ i, j ] is the gradient of x at [ i, j ]. dy [ i, j ] is the y gradient. M [ i, j ] is the gradient magnitude at that point. And theta [ i, j ] is the gradient direction.
And searching the local maximum value of the pixel point, comparing the front and rear gradient amplitude values along the gradient direction, if the local maximum value is the maximum value, retaining, and otherwise, performing non-maximum value inhibition. And when the suppressed gradient value is larger than a large threshold value, defining the gradient value as a strong edge point. The portions between the size thresholds are weak edge points. The portion smaller than the threshold is not an edge point, and its gradation value is 0.
The principles of binarization processing and perspective transformation are as follows:
and selecting an R channel of RGB, an S channel of HLS and a V channel of HSV in the color space as binaryzation reference channels according to the expression of the image in the color space and the effect in the edge extraction.
After binarization, significant target information can be completely extracted, but due to edge detection, similar information close to the target is inevitably detected. It is therefore desirable to convert an image from a single camera perspective to another perspective, i.e., a perspective transformation. The planar graph is firstly three-dimensionally transformed into a three-dimensional space through a transformation matrix, and then is transformed into a corresponding visual angle through perspective.
Performing parameter fitting on the third preprocessed image to obtain a stress-strain image according to the principle that:
the digital speckle technology is used for measuring fatigue durability stress-strain images, has high precision, and the coincidence degree of a stress-strain curve or deformation information obtained through reasonable parameter fitting and an actual measured value is ideal. The fitting parameters have maximum values in the template region, and accurate positions before and after deformation are searched through sub-pixel interpolation, so that relative displacement between corresponding points is determined, the strain is a derivative of the displacement, and a strain value can be obtained after conversion.
And taking out the images to be subjected to correlation operation from the speckle image series, selecting a module to be subjected to correlation operation in a specific area of the images, and performing correlation matching on each frame of speckle images by using the template so as to determine the detection robustness.
Optionally, the fatigue durability evaluating method in the above embodiment further includes:
obtaining a training sample;
training the initial stage of the target neural network model based on the training samples, and deriving and updating the weight and the bias term of the convolution kernel through a loss function to obtain the global optimal weight of the target neural network model.
Specifically, the loss function is an index for measuring the predicted value and the actual value of the model, and the commonly used loss function is a cross entropy loss function:
when the classification is correct, y is 1, and when the classification is incorrect, y is 0. a is the corresponding output probability.
When one-time back propagation is carried out and the parameter of the convolution kernel is updated, the weight w in the kernel of the convolution kernel is updatedjAnd (3) gradient calculation is carried out:
σ (z) is a Sigmoid function, and σ' (z) is σ (z) (1- σ (z)).
Similarly, the gradient formula of the bias recommendation in the convolution kernel is as follows:
the model training mainly refers to updating parameters of a convolution kernel, namely, derivation and updating are carried out on a weight and a bias term of the convolution kernel through a loss function:
wnα is the learning rate, ranging between (0, 1).
The selection of the learning rate is different along with the increase of the iteration times, a large learning rate is needed in the initial stage of model training to enable the model to be rapidly converged, the weight value of the model is close to the global optimum after multiple iterations, and a low learning rate is needed at the moment to prevent the model from missing the global optimum point. The selection of the learning rate often employs linear and exponential decay strategies.
Optionally, the fatigue durability evaluating method of the above embodiment further includes:
after the initial stage of the target neural network model is trained based on the training samples, the recall ratio and/or precision ratio of the training results are evaluated, and the optimal learning rate required by the model is obtained from the two.
Specifically, the recall ratio Re and the precision ratio Pr of the training results are evaluated:
the model prediction value is the probability of whether the object is a positive example, and when the predicted probability is greater than a certain threshold value, the model predicts the object as a positive example, otherwise, the model predicts the object as a negative example. When the threshold value is larger, the precision ratio of the model is larger, and the recall ratio is smaller. When the threshold is smaller, the precision of the model is smaller, but the recall is larger. Precision and recall are a pair of contradictory measures, the larger one, the smaller the other. The model therefore needs to find a balance between precision and recall.
The fatigue limit represents the bearing capacity of the material to the periodic stress/strain, and refers to the maximum stress/strain value when the material is not damaged after infinite stress cycles, which is also called as a endurance limit. In order to provide a component with sufficient reliability, it is necessary to ensure a sufficient safety margin between the strength of the component and the external load. The integral safety coefficient of the component in any fatigue failure mode is as follows:
γ=R/L;
r is the strength of the member. L is the safe life of the component.
The method of introducing the polynomial coefficient is to replace variables in the extreme state equation by nominal values, multiply each nominal value by a corresponding polynomial coefficient, and represent the functional function of the component by the nominal values and the polynomial coefficients, wherein the nominal values can be set as the mean value or other fractal values of the variables.
βTGiven a target reliability index. r isNIs the nominal value of the intensity, mean value is muR。sNThe mean value is the nominal value of the stresss,λ=μR/μs. d is 0.5 theta-0.4, theta is a design constant0/w)≥6,d0Is the outer diameter of the component and w is the wall thickness.
The relation function of the component system performance and the fatigue characteristics comprises a large number of uncertain parameters, and the parameter characteristics have different degrees of influence on the fatigue durability of the component. Usually, the corresponding relation between the design parameters and the application targets can be ensured in the design process of the component. Therefore, it is necessary to have a method for determining the parameters of the component designed to meet the fatigue durability requirements. For the problem of correlation between failure modes, in past research, a linear correlation coefficient is mostly used to describe the correlation degree between failure modes, when a performance function is linear or approximately linear, the coefficient can describe the correlation more accurately, and when the performance function is nonlinear, the accuracy degree of the coefficient needs to be further analyzed.
When the component has multiple failure modes, mutual independence assumption is carried out among the modes, namely, fatigue failure characteristics are defined as a series relation, so that the analysis difficulty can be reduced, and clear mapping relation between the component performance and the fatigue characteristics can be obtained quickly.
For the fatigue characteristics of the component with uncertain parameters, uncertain correlation and correlation degree between the parameters need to be considered in the process of designing and matching application. Therefore, the representation of the component parameters is characterized according to the fatigue durability requirements.
P is the probability of failure. gi(X1,X2,…,Xn) For the failure of the mouldRandom variable parameter in the formula. And N is the number of failure modes. C (P)1,Ph,…,PN) Is a Copula function.
When the number of failure modes is large, it takes a lot of calculation time to solve the reliability and reliability sensitivity of the components. Therefore, when analyzing the fatigue durability, the characteristic that the failure mode is less relative to the structural system is fully utilized, the analysis process is properly simplified, and the problem solving efficiency is improved.
The fatigue life of the transmission component is influenced by various factors, the parameters have obvious correlation, and the parameters interact together under most conditions until fatigue failure occurs, so that not only is the average stress influence the fatigue life, but also the frequency, the load retention time, the temperature and the like influence the fatigue life. The traditional fatigue test usually requires that the sample capacity is large enough to obtain comprehensive and accurate useful data, but for some transmission components with complex geometric shapes and processes and high processing precision requirements, the adoption of a full-life test undoubtedly requires the investment of a large amount of manpower, material resources and financial resources, and is obviously unrealistic. Therefore, reasonable acceleration tests and quantitative statistical methods are adopted, and the rule that the fatigue durability characteristics are reflected by limited samples can be fully utilized. The fatigue characteristic acceleration test is a test method for shortening the test period by increasing the test excitation condition under the condition of keeping the fatigue failure mechanism unchanged. The purposes of shortening the test time, improving the test efficiency and reducing the test cost are achieved.
A fatigue characteristic acceleration test method introduced into a frequency domain method comprises the following steps:
t is the fatigue life. And m is a material constant. SigmaRMSIn response to the stress level, (MPa). gRMSAt different excitation spectral magnitudes. K is a constant related to the intrinsic characteristics of the component,a is a constant relating to material properties only for a given excitation spectrum typeAnd (4) counting.
The embodiment realizes multi-system combined monitoring, various different sensing test devices are connected to the same test platform to perform system integrated control, long-time test monitoring, data acquisition, visual identification, thermal imaging detection and the like, comprehensive tests are convenient to realize, and test resources are fully utilized.
According to the intelligent prediction system, unattended intelligent prediction is realized, the intelligent prediction system for the fatigue durability test of the mechanical transmission component is used for carrying out test monitoring in an unattended or unattended long-term mode in an autonomous, network and even cloud control mode, abnormal phenomena such as fatigue, long-term fault-free limit and failure in the transmission component test can be found in time, and automatic alarm or preset fault treatment is carried out. The method has accurate operation, and avoids the defects of large labor intensity, low working efficiency, dispersed detection quality, single means, incapability of accurately and timely accessing detected data to a management information system and the like in a manual evaluation mode. Really plays the role of reducing the number of workers and improving the efficiency. The method can well ensure the evaluation accuracy and timeliness and promote the intelligent evaluation unattended process.
The method realizes the fatigue failure real-time detection of early fault feature perception, collects and simulates a large amount of early stress-strain cloud picture data sets, and extracts the fault micro-change, structural deformation and thermal imaging deformation images in the fatigue test process by using the deep learning neural network model with high real-time performance and high operation speed. The failure criterion can be judged in real time by using a deep learning method with high robustness to predict the failure.
Referring to fig. 5, the fatigue durability evaluation apparatus provided in this embodiment is applied to a mechanical transmission member, and includes:
the first obtaining module 31 is configured to obtain fatigue failure characteristic information and fatigue process characteristic information of the measured component, where the fatigue failure characteristic information is used to characterize a fatigue limit state of the measured component, and the fatigue process characteristic information is used to characterize a motion process state of the measured component before the fatigue limit state fails;
a second acquisition module 32, configured to acquire a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
the processing module 33 is configured to analyze and process the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
Optionally, the first obtaining module 31 includes:
the multi-influence parameter fatigue failure relation model module is used for constructing a multi-influence parameter fatigue failure relation model of the measured component;
and the fatigue failure characteristic information module is used for determining the characteristic information of the fatigue limit state of the tested component according to the multi-influence parameter fatigue failure relation model.
Optionally, the multiple influence parameter fatigue failure relation model includes a fatigue failure function linear approximation model and a fatigue failure function nonlinear approximation model, and the multiple influence parameter fatigue failure relation model module includes:
the first determining module is used for determining the quantitative relation of a single influence parameter of the measured object and the coupling association nonlinear relation of a plurality of influence parameters;
the fatigue failure function linear approximation model module is used for constructing a fatigue failure function linear approximation model of the measured component according to the quantitative relation of the single influence parameter;
and the fatigue failure function nonlinear approximate model module is used for constructing a fatigue failure function nonlinear model of the tested component according to the coupling association nonlinear relation of the plurality of influence parameters.
Optionally, the first obtaining module 31 includes:
the universal model module is used for constructing a universal model of the fatigue durability of the tested component;
and the second determination module is used for determining the fatigue process characteristic information of the tested component according to the fatigue durability universal model.
Optionally, the second obtaining module 32 includes:
the amplitude calculation module is used for identifying and reading the stress strain electronic speckle image by using a machine vision fuzzy algorithm based on a neural network to obtain stress and/or strain amplitude;
the calibration module is used for carrying out image distortion calibration processing on the stress strain electronic speckle image, correcting image distortion caused by the lens precision or the assembly process of the acquisition system and obtaining the calibrated stress strain electronic speckle image;
the color space transformation and edge extraction module is used for performing color space transformation on HSV and HLS in the calibrated stress-strain electronic speckle image and obtaining a graying processing result by adopting a weighted average method; performing edge extraction on the gray processing result by using a Canny operator to obtain an edge extraction result;
a binarization processing and perspective transformation module for selecting a target channel in a color space based on the graying processing result and the edge extraction result; determining the target channel as a binaryzation reference channel; performing binarization processing and perspective transformation on the stress strain electronic speckle image based on the binarized reference channel to obtain a transformation result;
and the parameter fitting module is used for performing parameter fitting based on the transformation result to obtain the reconstruction characteristic information of the stress-strain electronic speckle image and determining the reconstruction characteristic information as the fatigue process characteristic information.
Optionally, the fatigue durability evaluating apparatus further includes:
the training sample module is used for obtaining a training sample;
and the training module is used for training the initial stage of the target neural network model based on the training samples, and deriving and updating the weight and the bias term of the convolution kernel through a loss function to obtain the global optimal weight of the target neural network model.
Optionally, the fatigue durability evaluating apparatus further includes:
and the optimal learning rate module is used for evaluating the recall ratio and/or the precision ratio of the training result after the initial stage of the target neural network model is trained on the basis of the training samples, and obtaining the optimal learning rate required by the model from the recall ratio and/or the precision ratio.
Referring to fig. 6, an embodiment of the present invention further provides an electronic device 400, which includes a communication interface 401, a processor 402, a memory 403, and a bus 404, where the processor 402, the communication interface 401, and the memory 403 are connected by the bus 404; the memory 403 is used for storing a computer program that supports the processor 402 to execute the fatigue endurance evaluation method, and the processor 402 is configured to execute the program stored in the memory 403.
Optionally, an embodiment of the present invention further provides a computer-readable medium having a non-volatile program code executable by a processor, where the program code causes the processor to execute the fatigue durability evaluation method in the above embodiment.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A fatigue durability evaluation method is applied to a mechanical transmission component, and is characterized by comprising the following steps:
acquiring fatigue failure characteristic information and fatigue process characteristic information of a measured component, wherein the fatigue failure characteristic information is used for representing the fatigue limit state of the measured component, and the fatigue process characteristic information is used for representing the motion process state of the measured component before the measured component is in the fatigue limit state and fails;
acquiring a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
analyzing and processing the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
2. The method of claim 1, wherein obtaining fatigue failure characteristic information for the measured component comprises:
constructing a multi-influence parameter fatigue failure relation model of the measured component;
and determining characteristic information of the fatigue limit state of the tested component according to the multi-influence parameter fatigue failure relation model.
3. The method of claim 2, wherein the multiple influence parameter fatigue failure relationship model comprises a fatigue failure function linear approximation model and a fatigue failure function nonlinear approximation model, and constructing the multiple influence parameter fatigue failure relationship model of the measured component comprises:
determining the quantitative relation of a single influence parameter of the measured object and the coupling correlation nonlinear relation of a plurality of influence parameters;
constructing a fatigue failure function linear approximation model of the measured component according to the quantitative relation of the single influence parameter;
and constructing a fatigue failure function nonlinear model of the tested member according to the coupling association nonlinear relation of the plurality of influence parameters.
4. The method of claim 1, wherein obtaining fatigue failure characteristic information for the measured component comprises:
constructing a fatigue durability universal model of the tested component;
and determining the fatigue process characteristic information of the tested component according to the fatigue durability universal model.
5. The method of claim 1, wherein a stress-strain electronic speckle image of the measured member is acquired; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress-strain electronic speckle image comprises the following steps:
recognizing and reading the stress-strain electronic speckle image by using a machine vision fuzzy algorithm based on a neural network to obtain stress and/or strain amplitude;
carrying out image distortion calibration processing on the stress strain electronic speckle image, and correcting image distortion caused by the lens precision or the assembly process of the acquisition system to obtain a calibrated stress strain electronic speckle image;
performing color space transformation on HSV and HLS in the calibrated stress-strain electronic speckle image, and obtaining a graying processing result by adopting a weighted average method; performing edge extraction on the gray processing result by using a Canny operator to obtain an edge extraction result;
selecting a target channel in a color space based on the graying processing result and the edge extraction result; determining the target channel as a binaryzation reference channel; performing binarization processing and perspective transformation on the stress strain electronic speckle image based on the binarized reference channel to obtain a transformation result;
and performing parameter fitting based on the transformation result to obtain reconstruction characteristic information of the stress-strain electronic speckle image, and determining the reconstruction characteristic information as the fatigue process characteristic information.
6. The method of claim 1, further comprising:
obtaining a training sample;
training the initial stage of the target neural network model based on the training samples, and deriving and updating the weight and the bias term of the convolution kernel through a loss function to obtain the global optimal weight of the target neural network model.
7. The method of claim 6, further comprising:
after the initial stage of the target neural network model is trained based on the training samples, the recall ratio and/or precision ratio of the training results are evaluated, and the optimal learning rate required by the model is obtained from the two.
8. A fatigue durability evaluating device applied to a mechanical transmission component is characterized by comprising:
the device comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring fatigue failure characteristic information and fatigue process characteristic information of a measured component, the fatigue failure characteristic information is used for representing a fatigue limit state of the measured component, and the fatigue process characteristic information is used for representing a motion process state of the measured component before the measured component fails in the fatigue limit state;
the second acquisition module is used for acquiring a stress-strain electronic speckle image of the measured component; obtaining fatigue failure characteristic information and fatigue process characteristic information according to the stress strain electronic speckle image;
the processing module is used for analyzing and processing the fatigue failure characteristic information and the fatigue process characteristic information through a target neural network model to obtain a data prediction result of the measured component; and estimating the durability of the tested component according to the data prediction result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the method of any of the preceding claims 1 to 7 are implemented when the computer program is executed by the processor.
10. A computer-readable medium having non-volatile program code executable by a processor, wherein the program code causes the processor to perform the method of any of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911424498.3A CN111238927A (en) | 2019-12-30 | 2019-12-30 | Fatigue durability evaluation method and device, electronic equipment and computer readable medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911424498.3A CN111238927A (en) | 2019-12-30 | 2019-12-30 | Fatigue durability evaluation method and device, electronic equipment and computer readable medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111238927A true CN111238927A (en) | 2020-06-05 |
Family
ID=70870722
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911424498.3A Pending CN111238927A (en) | 2019-12-30 | 2019-12-30 | Fatigue durability evaluation method and device, electronic equipment and computer readable medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111238927A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115376612A (en) * | 2022-09-13 | 2022-11-22 | 郑州思昆生物工程有限公司 | Data evaluation method and device, electronic equipment and storage medium |
CN117554218A (en) * | 2024-01-12 | 2024-02-13 | 苏交科集团股份有限公司 | Straight asphalt pouring type steel bridge surface composite beam test piece fatigue test device and method |
-
2019
- 2019-12-30 CN CN201911424498.3A patent/CN111238927A/en active Pending
Non-Patent Citations (7)
Title |
---|
刘永胜: "《化学腐蚀下深部巷道围岩力学响应与安全控制》", 30 November 2007, 西南交通大学出版社 * |
张彬 等: "《图像复原优化算法》", 31 August 2019, 国防工业出版社 * |
徐成: "基于人工神经网络的铝合金搅拌摩擦焊接头疲劳寿命预测研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅰ辑》 * |
裔隽 等: "《Python机器学习实战》", 31 January 2018, 科学技术文献出版社 * |
贾星兰 等: "谱载下基于模糊Miner法则的疲劳寿命估算", 《北京航空航天大学学报》 * |
陈慧岩 等: "《无人驾驶车辆理论与设计》", 31 March 2018, 北京理工大学出版社 * |
黄丽 等: "《聚合物复合材料》", 31 January 2012, 中国轻工业出版社 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115376612A (en) * | 2022-09-13 | 2022-11-22 | 郑州思昆生物工程有限公司 | Data evaluation method and device, electronic equipment and storage medium |
CN115376612B (en) * | 2022-09-13 | 2023-10-13 | 郑州思昆生物工程有限公司 | Data evaluation method and device, electronic equipment and storage medium |
CN117554218A (en) * | 2024-01-12 | 2024-02-13 | 苏交科集团股份有限公司 | Straight asphalt pouring type steel bridge surface composite beam test piece fatigue test device and method |
CN117554218B (en) * | 2024-01-12 | 2024-03-22 | 苏交科集团股份有限公司 | Straight asphalt pouring type steel bridge surface composite beam test piece fatigue test device and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021000524A1 (en) | Hole protection cap detection method and apparatus, computer device and storage medium | |
CN109086876B (en) | Method and device for detecting running state of equipment, computer equipment and storage medium | |
CN110147781B (en) | Bridge vibration mode visual damage identification method based on machine learning | |
CN108956614A (en) | A kind of pit rope dynamic method for detection fault detection and device based on machine vision | |
CN113705788B (en) | Infrared image temperature estimation method and system based on full convolution neural network | |
CN116758491B (en) | Printing monitoring image analysis method and system applied to 3D printing | |
CN110796634A (en) | Dam state detection method and device, computer equipment and readable storage medium | |
CN113378988B (en) | Particle swarm algorithm-based robustness enhancement method and device for deep learning system | |
CN111238927A (en) | Fatigue durability evaluation method and device, electronic equipment and computer readable medium | |
CN110895814A (en) | Intelligent segmentation method for aero-engine hole detection image damage based on context coding network | |
CN109900363A (en) | A kind of object infrared measurement of temperature method and apparatus based on contours extract | |
Das et al. | Application of deep convolutional neural networks for automated and rapid identification and computation of crack statistics of thin cracks in strain hardening cementitious composites (SHCCs) | |
US11521120B2 (en) | Inspection apparatus and machine learning method | |
WO2019176988A1 (en) | Inspection system, identification system, and device for evaluating identification apparatus | |
Armingol et al. | Statistical pattern modeling in vision-based quality control systems | |
WO2023053029A1 (en) | Method for identifying and characterizing, by means of artificial intelligence, surface defects on an object and cracks on brake discs subjected to fatigue tests | |
CN111259494B (en) | Health monitoring and analyzing method for heavy machine equipment | |
Dunphy et al. | Autonomous crack detection approach for masonry structures using artificial intelligence | |
US20230214660A1 (en) | Hybrid training method for self-learining algorithms | |
CN117557574B (en) | Material parameter detection method and system based on image processing | |
Shin et al. | Visualization for explanation of deep learning-based defect detection model using class activation map | |
CN116754467B (en) | Evaluation method for ageing performance of natural rubber | |
CN114626162B (en) | Quantitative recognition method for loss degree of contact ball bearing | |
CN117150685B (en) | Fatigue life assessment system and method for fastening bolt | |
CN117474915B (en) | Abnormality detection method, electronic equipment and storage medium |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200605 |