Physical attribute parameter calibration method, system and equipment of composite material plate and storage medium
Through deterministic update of simulation signals and experimental signals and Bayesian inferred simulation, the calibration of physical attribute parameters of composite material plates is optimized, which solves the problem of high calculation costs in the prior art, improves calibration efficiency and accuracy, and enhances fatigue Status prediction.
Patent Information
- Application Number
- CN202411099146.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art has high computational cost in lamb wave propagation calibration of composite material plates, resulting in long time and high computing resource consumption, affecting the efficiency of fatigue life prediction.
The simulation signal and experimental signal are used for deterministic updates, combined with Bayesian inference and Monte Carlo simulation, iterative random updates are performed through the proxy model to optimize the calibration process of physical attribute parameters.
The efficiency and accuracy of physical attribute parameters calibration of composite material plates is improved, the calculation cost is reduced, and the accuracy of fatigue state prediction is enhanced.
Smart Images

Figure CN119092016A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of structural nondestructive testing, and in particular to a physical property parameter calibration method, system, equipment and storage medium for a composite material plate. Background Art
[0002] Composite plates are plates made of two or more materials, including composite metal plates and composite non-metallic plates. In recent years, composite materials have been widely used in aerospace and other fields due to their excellent mechanical properties such as high specific strength and specific stiffness. Composite materials suffer fatigue damage during use, such as matrix cracks, delamination, and fiber breakage. These damages interact with each other, accumulate and expand continuously, and easily lead to damage to composite plates. Therefore, fatigue life prediction of composite materials is very important. Generally speaking, the fatigue degree of composite plates is evaluated by comparing the difference between the physical property parameters of the composite materials at the time of leaving the factory (i.e., the specification physical property parameters) and the physical property parameters at the current moment. The greater the difference, the higher the fatigue degree is considered.
[0003] Lamb waves, also known as Lamb waves, are ultrasound-guided elastics that are widely used to predict physical property parameters in structural health monitoring (SHM) of composite plates. Semi-empirical methods and numerical methods can usually be used to guide the simulation of Lamb wave propagation in composite plate structures. In numerical methods, the finite element method (FE) and Bayesian updating are often used to calibrate and express the propagation of Lamb waves in complex structures. Specifically, the local interaction simulation approach (LISA) and the wavelet finite element method (WFE) are used to apply Bayesian inference to composite structures excited by Lamb waves to estimate the current structural state of the composite plate. However, there is a problem that the Bayesian inference process has a high computational cost, which leads to a long time and requires a lot of computing resources. Summary of the invention
[0004] In order to reduce the computational cost when calibrating Lamb wave propagation on a composite material plate and improve the efficiency of calibrating physical property parameters of the composite material plate, a method, system, device and storage medium for calibrating physical property parameters of a composite material plate are provided.
[0005] In order to achieve the above-mentioned object of the present invention, according to a first aspect of the present invention, the present invention provides a method for calibrating physical property parameters of a composite material plate.
[0006] The present invention provides a method for calibrating physical property parameters of a composite material plate, comprising:
[0007] Acquiring initial information of input parameters of the composite material plate, wherein the input parameters include a number of physical properties capable of characterizing mechanical properties of the composite material plate;
[0008] Inputting the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain a simulation signal;
[0009] Collect Lamb wave signals on the composite material plate in real time as experimental signals;
[0010] The initial information of the input parameters is deterministically updated using simulation signals and experimental signals to obtain the optimal property values of each physical property parameter in the input parameters;
[0011] The optimal property value of each physical property parameter is input into the proxy model, and the proxy model uses Bayesian inference and Monte Carlo simulation to iteratively and randomly update the physical property parameters to obtain the predicted values of the physical property parameters of the current composite material plate.
[0012] Optionally, the physical property parameters include Young's modulus, Poisson's ratio, shear modulus and density value.
[0013] Optionally, the deterministic updating of initial information of input parameters by using simulation signals and experimental signals to obtain optimal property values of various physical property parameters in the input parameters includes:
[0014] Calculate the residual between the experimental signal and the simulated signal;
[0015] Determining the optimal property value of each physical property parameter according to the residual;
[0016] The residual includes a first residual, a second residual and a third residual,
[0017] The first residual is used to quantify the amplitude and phase difference between the simulation signal and the experimental signal. The first residual is denoted as R NORM ,
[0018] The calculation formula for the first residual is:
[0019] Where x(t) is the experimental signal, y(t) is the simulation signal, t1 and t2 are two different moments in the signal time domain diagram, t1 is the sampling start time, and t2 is the sampling end time;
[0020] The second residual is denoted as RXCOR , the calculation formula of the second residual is R XCOR =1-r xy ,
[0021] Among them, r xy is the cross-correlation function value between the experimental signal and the simulation signal;
[0022] The third residual is used to represent the difference between the power spectrum of the simulated signal and the power spectrum of the experimental signal. The third residual is denoted as R PSD ,
[0023] The calculation formula for the third residual is:
[0024] Where X(f) is the power spectral density of the experimental signal, Y(f) is the power spectral density of the simulated signal, and f1 and f2 refer to two different frequencies in the signal spectrum.
[0025] The residual between the experimental signal and the simulated signal is denoted as R TOTAL , then the residual expression is R TOTAL =k1R NORM +k2R XCOR +k3R PSD ,
[0026] Where k1 is the weight of the first residual, k2 is the weight of the second residual, and k3 is the weight of the third residual.
[0027] Optionally, determining the optimal property value of each physical property parameter according to the residual includes:
[0028] Disturb each physical property parameter separately, and change the value of each physical property parameter until the residual value converges;
[0029] When the residual value converges, it is determined that the current physical attribute parameter value of each physical parameter is the optimal attribute value.
[0030] Optionally, after the initial information of the input parameters is deterministically updated by using the simulation signal and the experimental signal, and before the optimal property value of the physical property parameter is input into the proxy model, the method further comprises:
[0031] Sensitivity analysis is performed on each physical property to evaluate the impact of each physical property on the behavior of the Lamb wave finite element model and determine the key parameters among the input parameters.
[0032] Optionally, the expression of the output response of the Lamb wave finite element model is Y = M (Θ),
[0033] Where θ represents the input parameters, specifically the physical properties; θ=(θ1,θ2,…,θ k) is a vector containing k input variables, where k is the index of the input variable;
[0034] The sensitivity analysis of each physical property parameter is performed to evaluate the influence of each physical property on the behavior of the Lamb wave finite element model, and the key parameters in the input parameters are determined to include:
[0035] The Sobol index is used to perform sensitivity analysis on the output response Y of the Lamb wave finite element model in different dimensions;
[0036] The Sobol index is analyzed by Monte Carlo simulation to evaluate the influence of each physical property on the behavior of the Lamb wave finite element model and determine the key parameters among the input parameters;
[0037] The decomposition expression of the Lamb wave finite element model output response in different dimensions is:
[0038]
[0039] The constant Μ0 is the mean value of the output response of the Lamb wave finite element model, M i (θ i ) represents the conditional mean of input parameter i, M ij (θ i ,θ j ) represents the conditional mean of input parameter ij when input parameter i and input parameter j (i≠j) interact with each other; M 1…k (θ1…θ k ) represents the conditional mean of input parameters 1…k when k input parameters interact with each other;
[0040] M i (θ i ) is expressed as ~i means that the parameter θ is excluded i ;
[0041] M ij (θ i ,θ j ) is expressed as
[0042] ~ij means that the parameter θ is excluded ij ;
[0043] By calculating the first-order Sobol index of input parameters and output responses, the sensitivity measurement results of individual physical properties are obtained;
[0044] The expression of the first-order Sobol index is Var represents variance;
[0045] By calculating the second-order Sobol index of input parameters and output responses, we can obtain the sensitivity measurement results when two physical properties interact with each other.
[0046] The expression of the second-order Sobol index is Key Parameters Key Parameters.
[0047] Optionally, the proxy model randomly updates the physical property parameters using Bayesian inference and Monte Carlo simulation, including:
[0048] Determine the prior distribution π(Θ) of key parameters among the input parameters based on a deterministic update process;
[0049] Determine the posterior distribution density function of the input parameter Θ according to the prior distribution π(Θ),
[0050] The calculation formula of the posterior distribution density function is:
[0051] in, is the observed random data, is the conditional likelihood function, is the marginal likelihood function;
[0052] The Monte Carlo sampling algorithm is used to sample the posterior distribution density function to generate a series of random variables;
[0053] Based on the generated random variables, simulation experiments are carried out to obtain experimental results, and parameters are estimated according to the experimental results to obtain the predicted values of the physical property parameters of the current composite material plate.
[0054] In order to solve the above problems, the present invention also provides a physical property parameter calibration system for a composite material plate, the system comprising:
[0055] An acquisition module, used to acquire initial information of input parameters of the composite material plate, wherein the input parameters include a number of physical properties that can characterize the mechanical properties of the composite material plate;
[0056] A simulation module is used to input the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain a simulation signal;
[0057] An acquisition module is used to acquire Lamb wave signals on the composite material plate in real time as experimental signals;
[0058] A deterministic updating module is used to deterministically update the initial information of the input parameters using simulation signals and experimental signals to obtain the optimal property value of each physical property parameter in the input parameters;
[0059] The random update module is used to input the optimal property value of each physical property parameter into the proxy model. The proxy model uses Bayesian inference and Monte Carlo simulation to iteratively and randomly update the physical property parameters to obtain the predicted values of the physical property parameters of the current composite material plate.
[0060] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:
[0061] at least one processor; and,
[0062] a memory communicatively connected to the at least one processor; wherein,
[0063] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the above-mentioned method for calibrating physical property parameters of the composite material plate.
[0064] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one computer program is stored. The at least one computer program is executed by a processor in an electronic device to implement the above-mentioned method for calibrating physical property parameters of a composite material plate.
[0065] In summary, this application includes the following beneficial technical effects:
[0066] When performing random calibration of Lamb wave propagation in a composite structure, the initial information of the input parameters is deterministically updated using the simulation signal and the experimental signal, and then the physical property parameters of the input parameters are randomly updated using Bayesian inference and Monte Carlo, so as to improve the accuracy of the Bayesian calibration results of Lamb wave propagation in the composite structure, so as to make the analysis results of the predicted current physical property parameters of the composite material plate more accurate, and compare the predicted physical property parameters of the composite material plate at the current moment with the physical property parameters of the composite material plate when it leaves the factory, so as to evaluate the current fatigue degree of the composite material plate, and improve the accuracy of the prediction results of the current fatigue state of the composite material plate;
[0067] Using a proxy model to assist the Lamb wave finite element model in Bayesian inference can improve the calculation speed of the Lamb wave finite element model, reduce the time required for Monte Carlo simulation rounds, reduce the computational cost when calibrating Lamb wave propagation on composite plates, and improve the calibration efficiency of the physical property parameters of composite plates. BRIEF DESCRIPTION OF THE DRAWINGS
[0068] Figure 1 A schematic flow chart of a method for calibrating physical property parameters of a composite material plate provided in one embodiment of the present invention;
[0069] Figure 2 A schematic diagram of the steps of a method for calibrating physical property parameters of a composite material plate provided by an embodiment of the present invention;
[0070] Figure 3 A schematic diagram of a process for deterministically updating input parameter information using simulation signals and experimental signals provided by an embodiment of the present invention;
[0071] Figure 4 A step diagram of a method for calibrating physical property parameters of a composite material plate provided in a preferred embodiment of the present invention;
[0072] Figure 5 A schematic diagram of a proxy model provided by an embodiment of the present invention using Bayesian inference and Monte Carlo simulation to perform random updating steps on input parameter information;
[0073] Figure 6 A diagram showing the structure of a multi-layer perceptron neural network according to an embodiment of the present invention;
[0074] Figure 7 A functional module diagram of a physical property parameter calibration system for a composite material plate provided by an embodiment of the present invention;
[0075] Figure 8 A schematic diagram of the structure of an electronic device for implementing the physical property parameter calibration method of a composite material plate provided in one embodiment of the present invention.
[0076] Figure numerals: 71, acquisition module; 72, simulation module; 73, acquisition module; 74, deterministic update module; 75, random update module; 10, processor; 11, memory; 12, communication bus; 13, communication interface.
[0077] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0078] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0079] In the description of the present invention, it is necessary to understand that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.
[0080] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.
[0081] Reference Figure 1 and Figure 2 , is a method for calibrating physical property parameters of a composite material plate provided by an embodiment of the present invention. In this embodiment, the method for calibrating physical property parameters of a composite material plate includes:
[0082] S1. Obtaining initial information of input parameters of the composite material plate.
[0083] The input parameters include several physical property parameters (i.e., specification physical property parameters) that can characterize the mechanical properties of the composite material plate. The initial information of the input parameters is the physical property parameters of the current composite material plate when it leaves the factory. Specifically, the physical property parameters include Young's modulus, Poisson's ratio, shear modulus and density values. Among them, Young's modulus is used to describe the shape change of the composite material plate after being subjected to force, especially the stretching or compression in one-dimensional direction. The shear modulus is used to describe the ability of the composite material plate to resist deformation when subjected to shear force. The Poisson's ratio is an elastic constant that reflects the lateral deformation of the material and is used to describe the deformation mode of the material when subjected to external force. The density of the composite material plate refers to the mass contained in a unit volume of the plate.
[0084] S2. Input the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain a simulation signal.
[0085] In this embodiment, the input parameter is the input value of the Lamb wave finite element model. The Lamb wave finite element model is a model established based on the finite element analysis method. It is composed of a group of unit assemblies that are connected only at nodes, transmit forces only at nodes, and are constrained only at nodes. The Lamb wave finite element model determines the current physical state of the composite material plate by calculating the dynamic stiffness matrix and the actually measured Lamb wave signal, and can detect and evaluate the damage of the composite material plate.
[0086] In this embodiment, each layer of the composite plate is considered to be an orthotropic material. The input parameter information includes two Young's moduli in two perpendicular directions on the plane of the composite plate, Poisson's ratio, two shear moduli inside and outside the plane of the composite plate, and density; the two Young's moduli are respectively denoted as E1 and E2, and the Poisson's ratio is denoted as v 12 , and the two shear moduli are denoted as G 12 and G 23 , density is recorded as ρ; establish an initial search range for the above physical parameters according to the composite material standard requirements. The composite material standard requirements can adopt any existing international or domestic standards, or can be formulated according to actual needs. This embodiment does not limit it. 12 , G 12 , G 23 and ρ are taken as input parameters into the Lamb wave finite element model for modeling and simulation to obtain the simulation signal.
[0087] S3. Collect Lamb wave signals on the composite material plate in real time as experimental signals.
[0088] Specifically, the Lamb wave signal on the composite material board can be collected by a power amplifier or a bandpass filter, and the measurement data can be transmitted to the monitoring system by wireless or wired means.
[0089] S4. Use simulation signals and experimental signals to deterministically update the initial information of the input parameters to obtain the optimal property value of each physical property parameter in the input parameters.
[0090] Reference Figure 3 In this embodiment, the specific implementation steps of step S4 are:
[0091] S41, calculating the residual between the experimental signal and the simulation signal;
[0092] In this embodiment, the residual between the experimental signal and the simulation signal is calculated using an improved least square method, and the influence of each input parameter on the loss function of the deterministically updated Lamb wave finite element model is linearized.
[0093] The residual includes a first residual, a second residual and a third residual, and the first residual, the second residual and the third residual are used to measure the difference between the experimental and simulation signals.
[0094] The first residual is used to evaluate the square norm of the difference between the experimental and simulated signals. The first residual can quantify the amplitude and phase differences between the simulated and experimental signals. The first residual is recorded as R NORM , then the first residual R NORM The calculation formula is
[0095]
[0096] Wherein, x(t) is the experimental signal, y(t) is the simulation signal, t1 is the sampling start time, t2 is the sampling end time, t1 is the lower limit of the integral in the first residual calculation formula, and t2 is the upper limit of the integral in the first residual calculation formula;
[0097] The second residual is denoted as R XCOR , the second residual R XCOR The calculation formula of the second residual based on the cross-correlation between the experimental signal and the simulated signal is R XCOR =1-r xy ,
[0098] Among them, r xy It is the cross-correlation function value between the experimental signal and the simulation signal, which indicates the similarity between the experimental signal and the simulation signal and is only sensitive to changes in signal shape and frequency. In signal processing, the cross-correlation function is usually represented as the convolution between two signals. Any method in the prior art can be used to calculate the cross-correlation function between the experimental signal and the simulation signal. It is sufficient to obtain the cross-correlation function between the experimental signal and the simulation signal. This embodiment does not impose any limitation.
[0099] The third residual is used to represent the difference between the power spectrum of the simulated signal and the power spectrum of the experimental signal, which can detect the delay and frequency offset in the signal; the third residual is recorded as R PSD ,
[0100] The calculation formula for the third residual is:
[0101] Where X(f) is the power spectral density of the experimental signal, Y(f) is the power spectral density of the simulated signal, f1 and f2 refer to two different frequencies in the spectrum of the experimental signal (simulated signal), f1 is the lower limit of the integral in the third residual calculation formula, and f2 is the upper limit of the integral in the third residual calculation formula;
[0102] The residual between the experimental signal and the simulated signal is denoted as R TOTAL The expression of the residual is R TOTAL =k1R NORM +k2R XCOR +k3R PSD ,
[0103] Where k1 is the weight of the first residual, k2 is the weight of the second residual, and k3 is the weight of the third residual. In actual use, the values of k1, k2, and k3 can be adjusted according to actual conditions to control the relative importance of each indicator in the least squares result.
[0104] S42, determining the optimal attribute value of each physical attribute parameter according to the residual;
[0105] Determining the optimal attribute values of each physical attribute parameter based on the residual includes:
[0106] Calculate the partial derivative of the residual, perturb each physical property parameter in the input parameter information respectively, and change the numerical value of each physical property parameter until the residual value converges; when the residual value converges, it is determined that the current physical property parameter value of each physical parameter is the optimal property value.
[0107] In some examples of this embodiment, the disturbance value of the physical property parameter is initially set to 5%, and the disturbance value refers to the amplitude of the numerical adjustment of the physical property parameter of the composite material plate; during the optimization process, the disturbance value of the physical property parameter is gradually reduced to reduce system instability.
[0108] S5. The optimal property value of each physical property parameter is input into the proxy model. The proxy model uses Bayesian inference and Monte Carlo simulation to iteratively and randomly update the physical property parameters to obtain the predicted value of the physical property parameter of the current composite material plate.
[0109] Reference Figure 4 In a preferred implementation of this embodiment, after the physical property parameters are deterministically updated using the simulation signal and the experimental signal, and before the optimal property values of the physical property parameters are input into the proxy model, the physical property parameter calibration method of the composite material plate further includes:
[0110] S31. Conduct sensitivity analysis on each physical property, evaluate the impact of each physical property on the behavior of the Lamb wave finite element model, and determine the key parameters among the input parameters.
[0111] Specifically, the behavior of the Lamb wave finite element model refers to the response output signal of the model. The expression of the Lamb wave finite element model output response is Y = M (Θ),
[0112] Where θ represents the input parameters, specifically the physical properties of the composite material plate; θ = (θ1, θ2, …, θ k ) is a vector containing k input variables, k is the index of the input variable, and the input parameters are uniformly distributed in the support interval π(Θ)∈U(0,1).
[0113] A sensitivity analysis was performed on each physical property to evaluate the impact of each physical property on the behavior of the Lamb wave finite element model and to determine the key parameters in the input parameters, including:
[0114] The Sobol index is used to perform sensitivity analysis on the output response Y in different dimensions;
[0115] The Sobol index is analyzed through Monte Carlo simulation to evaluate the influence of each physical attribute on the behavior of the Lamb wave finite element model, and the influence of the interaction between different physical attributes on the behavior of the Lamb wave finite element model is evaluated to determine the key parameters among the input parameters;
[0116] The decomposition expression of the Lamb wave finite element model output response in different dimensions is:
[0117]
[0118] The constant M0 is the mean value E(Y) of the output response of the Lamb wave finite element model, M i (θ i ) represents the conditional mean of input parameter i, M ij (θ i ,θ j ) represents the conditional mean of input parameter ij when input parameter i and input parameter j (i≠j) interact with each other; M 1…k (θ1…θ k ) represents the conditional mean of input parameters 1…k when k input parameters interact with each other;
[0119] M i (θ i ) is expressed as ~i means that the parameter θ is excluded i ;
[0120] M ij (θ i ,θ j ) is expressed as
[0121] ~ij means that the parameter θ is excluded ij ;
[0122] Since the output response Y is also a random variable, the conditional expected variance can be viewed as a summary measure of sensitivity. Therefore,
[0123] The sensitivity measures of individual physical properties are obtained by calculating the first-order Sobol indices of input parameters and output responses, which are used to quantify the additive effect of each input parameter on the total variance. The first-order index represents the main contribution of each input parameter to the model output variance.
[0124] The expression of the first-order Sobol index is Var represents variance;
[0125] Similarly, by calculating the second-order Sobol index of input parameters and output responses, the sensitivity measurement results when two physical properties interact with each other are obtained;
[0126] The expression of the second-order Sobol index is
[0127] In the preferred implementation of this embodiment, the first-order Sobol index is analyzed by Monte Carlo simulation to evaluate the influence of each input parameter on the model parameters of the Lamb wave finite element model, so as to determine the key parameters that significantly affect the response results of the Lamb wave finite element model among all the input parameters; Figure 1 In the specific example of this embodiment, the key parameters in the input parameter information are Young's modulus E1 and shear modulus G 12 .
[0128] The surrogate model uses Bayesian inference and Markov chain-based Monte Carlo simulation to randomly update the values of key parameters and complete the calibration of Lamb wave propagation on the composite plate.
[0129] There are inherent variations in physical systems, such as geometry, environmental conditions, material properties, etc. Therefore, even the most complete data set cannot eliminate this uncertainty, the so-called irreducible uncertainty. In order to include this type of variability in the model parameters of the computational surrogate model, the present invention adopts the Bayesian paradigm as a statistical inference tool for the Lamb wave finite element model calibration problem.
[0130] Reference Figure 5 Specifically, the agent model uses Bayesian inference and Monte Carlo simulation to randomly update the input parameter information including:
[0131] S51, determining a prior distribution π(Θ) of key parameters in the input parameters based on a deterministic update process;
[0132] The prior distribution of the input parameters can be determined by a variety of methods such as maximum likelihood estimation method, moment method, subjective judgment and experience, selection of conjugate prior family, and setting of hyperparameters. The method of determining the prior distribution can be determined according to the specific application scenario and data characteristics, and this embodiment does not limit it. In this embodiment, the range of the prior distribution of the key parameters is set to: a value of ±20% of the optimal value of the attribute updated deterministically.
[0133] S52. Determine the posterior distribution density function of the input parameter Θ according to the prior distribution π(Θ). The calculation formula of the posterior distribution density function is:
[0134] in, is the observed random data, is the conditional likelihood function, is the marginal likelihood function;
[0135] The relationship between the posterior distribution density function and the conditional likelihood function is: “∝” means proportional to, that is, the updated posterior probability density function is proportional to the likelihood distribution.
[0136] Whether the current physical property parameters of the composite material plate are accurate is judged according to the posterior distribution density function, so as to obtain the predicted value of the final physical property parameters of the composite material plate; the predicted value of the final physical property parameters is compared with the physical property parameters of the current composite material plate when it leaves the factory to obtain the current damage condition of the composite material plate.
[0137] S53, using the Monte Carlo sampling (Metropolis-Hastings) algorithm to sample the posterior distribution density function to generate a series of random variables;
[0138] S54. Perform simulation experiments based on the generated random variables to obtain experimental results, and perform parameter estimation based on the experimental results to obtain predicted values of the physical property parameters of the current composite material plate.
[0139] After obtaining the predicted values of the current physical property parameters of the composite material plate, the predicted values of the current physical property parameters of the composite material plate are compared with the physical property parameters of the composite material plate when it leaves the factory to evaluate the current fatigue degree of the composite material plate and realize the prediction of the current fatigue state of the composite material plate.
[0140] We can use the normal distribution to approximate the output response predicted by the Lamb wave finite element model, assuming X~N(μ,σ 2 ), μ represents the mean, σ 2 represents variance; if D M (Θ) is subject to zero mean and variance The additive nature of the Gaussian noise is irrelevant, so the conditional likelihood function is The analytical expression of is:
[0141]
[0142] Among them, D M (Θ) represents the sample distribution drawn in the Monte Carlo simulation, that is, the expectation of the output using Monte Carlo sampling; express The transpose of , the superscript T indicates the transpose operation; Indicates D M The variance of (Θ).
[0143] The Monte Carlo sampling algorithm based on Markov chain is used to sample the posterior distribution density function to generate a series of random variables; the random variable Θ is restricted to the interval [Θ min ,Θ max ], the random variable Θ is symmetrically normalized to obtain Θ'=(1-x)Θ min +xΘmax ; where Θ' is the variable after symmetric normalization of the random variable Θ; x is a random number in the range [0,1], a candidate sample generated from the posterior distribution; it should be noted that x is a random number with a standard deviation of σ p The normal distribution of the sample is sampled by the hyperparameter σ p Controls the step size of the random walk during parameter search during Bayesian inference to prevent the Markov chain from becoming static while allowing proper exploration of parameter space.
[0144] In the Monte Carlo algorithm, if a candidate sample is accepted, it is because the transfer from the current sample to the candidate sample increases the probability value of the target distribution. This acceptance-rejection mechanism ensures that the sampling process can explore the global space of the target distribution; the variance of the likelihood function is defined through multiple fitting algorithm iterations to ensure that the acceptance rate of candidate samples is stable between 40% and 50%. After multiple iterations, after appropriately adjusting the Markov chain parameters such as variance and step size in each round, the Monte Carlo sampling algorithm converges to an optimal region, and the Monte Carlo sampling algorithm continues to sample from the posterior distribution within this stable region.
[0145] Since Monte Carlo simulation is very time-consuming, in the preferred implementation of this embodiment, a proxy model is used to replace the finite element model in the random walk. There are many techniques to choose from for proxy modeling, such as Gaussian process, PCE (Polynomial Chaos Expansion) or ANN (Artificial Neural Network).
[0146] This application chooses to use ANN neural network to work, using a multi-layer perceptron neural network (Multi-Layer Perceptron, referred to as MLP) as the interpolation model of the parameter search space in the Bayesian inference process.
[0147] The structure of a multilayer perceptron neural network is as follows Figure 6 As shown, the input layer of the agent model contains two neurons, which represent the random variables E1 and G of two key parameters respectively. 12 ; The output layer has 140 neurons, representing the time series output. The number of hidden layers is defined as two, with N2 and N3 neurons in the first and second hidden layers, respectively.
[0148] The proxy model can adopt an existing neural network model. In some examples of this embodiment, before using the proxy model, the proxy model can also be trained using a Stochastic Gradient Descent with Momentum (SGDM) algorithm, and the network loss on the validation data is used as the criterion for stopping training. The difference between the proxy model output and the full-order finite element model is the error metric that guides the learning process. The training algorithm periodically evaluates the validation data, and if the loss on the validation data no longer decreases or begins to increase, the training process is stopped.
[0149] The performance of the proxy model is affected by its hyperparameters, such as the configuration of neurons in the hidden layer, the activation function, the initial learning rate, and the amount of available training data. The number of neurons in the hidden layer (i.e., the values of N2 and N3) is defined by performing an exploratory search within 50 to 300 neurons in each layer. It has been experimentally verified that in the preferred embodiment of this embodiment, 50 neurons are used in the second hidden layer and 200 neurons are used in the third hidden layer; the number of neurons in each hidden layer can be adjusted in actual use, and this embodiment does not impose any restrictions.
[0150] Reference Figure 7 Based on the same inventive concept, an embodiment of the present invention provides a physical property parameter calibration system for a composite material plate.
[0151] The physical property parameter calibration system of a composite material plate of the present invention can be loaded in an electronic device. According to the functions to be implemented, the physical property parameter calibration system of a composite material plate includes an acquisition module, a simulation module, a collection module, a deterministic update module and a random update module.
[0152] The acquisition module can obtain the initial information of the input parameters of the composite material plate, and the input parameters include several physical properties that can characterize the mechanical properties of the composite material plate; the simulation module can input the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain the simulation signal; the acquisition module can collect the Lamb wave signal on the composite material plate in real time as the experimental signal;
[0153] The deterministic update module can use simulation signals and experimental signals to deterministically update the initial information of the input parameters to obtain the optimal property value of each physical property parameter in the input parameters; the random update module can input the optimal property value of each physical property parameter into the proxy model, and the proxy model uses Bayesian inference and Monte Carlo simulation to iteratively randomly update the physical property parameters to obtain the predicted value of the physical property parameter of the current composite material plate. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.
[0154] The various variations and specific examples of the physical property parameter calibration method for composite panels provided in the above-mentioned embodiments are also applicable to the physical property parameter calibration system for composite panels of the present embodiment. Through the above-mentioned detailed description of the physical property parameter calibration method for composite panels, those skilled in the art can clearly understand the implementation method of the physical property parameter calibration system for composite panels of the present embodiment. For the sake of brevity of the specification, it will not be described in detail here.
[0155] The present application also discloses an electronic device, as shown in the figure, which is a structural schematic diagram of an electronic device for a method for calibrating physical property parameters of a composite material plate provided by an embodiment of the present invention. The electronic device may include at least one processor 10, a memory 11 connected to at least one processor for communication, a communication bus 12, and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a program for the method for calibrating physical property parameters of a composite material plate.
[0156] Among them, the processor 10 can be composed of an integrated circuit in some embodiments, for example, it can be composed of a single packaged integrated circuit, or it can be composed of multiple integrated circuits with the same function or different functions, including one or more central processing units (CPU), microprocessors, digital processing chips, graphics processors and various control chips. The processor 10 is the control core (Control Unit) of the electronic device, which uses various interfaces and lines to connect various components of the entire electronic device, and executes various functions of the electronic device and processes data by running or executing programs or modules stored in the memory 11 (for example, a method for calibrating physical property parameters of a composite material board, etc.), and calling data stored in the memory 11.
[0157] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of an electronic device, such as a mobile hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device, such as the code of the program of the physical property parameter calibration method of the composite material plate, but also can be used to temporarily store data that has been output or is to be output.
[0158] The communication bus 12 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.
[0159] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.
[0160] Figure 8 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 8The structure shown does not constitute a limitation on the electronic device, and may include fewer or more components than shown, or combine certain components, or arrange the components differently. For example, although not shown, the electronic device may also include a power supply (such as a battery) for supplying power to each component. Preferably, the power supply may be logically connected to at least one processor 10 through a power management device, so that functions such as charging management, discharging management, and power consumption management are implemented through the power management device. The power supply may also include any components such as one or more DC or AC power supplies, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, etc. The electronic device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.
[0161] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited by this structure.
[0162] Furthermore, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile.
[0163] The present application provides a computer-readable storage medium, for example, including: any entity or device, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM) that can carry the computer program code. The computer-readable storage medium stores a computer program that can be loaded by a processor and execute the physical property parameter calibration method of the composite material plate of the above embodiment.
[0164] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", "an implementation", "a preferred implementation" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0165] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.
Claims
1. A method for calibrating physical property parameters of a composite material plate, characterized in that: Acquiring initial information of input parameters of the composite material plate, wherein the input parameters include a number of physical property parameters capable of characterizing mechanical properties of the composite material plate; Inputting the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain a simulation signal; Collect Lamb wave signals on the composite material plate in real time as experimental signals; The initial information of the input parameters is deterministically updated using simulation signals and experimental signals to obtain the optimal property values of each physical property parameter in the input parameters; The optimal property value of each physical property parameter is input into the proxy model, and the proxy model uses Bayesian inference and Monte Carlo simulation to iteratively and randomly update the physical property parameters to obtain the predicted values of the physical property parameters of the current composite material plate.
2. The method for calibrating physical property parameters of a composite material plate according to claim 1, characterized in that: The physical property parameters include Young's modulus, Poisson's ratio, shear modulus and density values.
3. The method for calibrating physical property parameters of a composite material plate according to claim 1, characterized in that: The method of deterministically updating the initial information of the input parameters using the simulation signal and the experimental signal to obtain the optimal property value of each physical property parameter in the input parameters includes: Calculate the residual between the experimental signal and the simulated signal; Determining the optimal property value of each physical property parameter according to the residual; The residual includes a first residual, a second residual and a third residual, The first residual is used to quantify the amplitude and phase difference between the simulation signal and the experimental signal. The first residual is denoted as R NORM , The calculation formula for the first residual is: Among them, x(t) is the experimental signal, y(t) is the simulation signal, t1 is the sampling start time, and t2 is the sampling end time; The second residual is denoted as R XCOR , the calculation formula of the second residual is R XCOR =1-r xy , Among them, r xy is the cross-correlation function value between the experimental signal and the simulation signal; The third residual is used to represent the difference between the power spectrum of the simulated signal and the power spectrum of the experimental signal. The third residual is denoted as R PSD , The calculation formula for the third residual is: Where X(f) is the power spectral density of the experimental signal, Y(f) is the power spectral density of the simulated signal, and f1 and f2 refer to two different frequencies in the signal spectrum. The residual between the experimental signal and the simulated signal is denoted as R TOTAL , then the expression of the residual is For R TOTAL =k1R NORM +k2R XCOR +k3R PSD , Where k1 is the weight of the first residual, k2 is the weight of the second residual, and k3 is the weight of the third residual.
4. The method for calibrating physical property parameters of a composite material plate according to claim 3, characterized in that: Determining the optimal attribute value of each physical attribute parameter according to the residual includes: Disturb each physical property parameter separately, and change the value of each physical property parameter until the residual value converges; When the residual value converges, it is determined that the current physical attribute parameter value of each physical parameter is the optimal attribute value.
5. The method for calibrating physical property parameters of a composite material plate according to any one of claims 1 to 4, characterized in that: After the initial information of the input parameters is deterministically updated by using the simulation signal and the experimental signal, and before the optimal property value of the physical property parameter is input into the proxy model, the method further includes: Sensitivity analysis is performed on various physical property parameters to evaluate the impact of various physical properties on the behavior of the Lamb wave finite element model and determine the key parameters among the input parameters.
6. The method for calibrating physical property parameters of a composite material plate according to claim 5, characterized in that: The expression of the output response of the Lamb wave finite element model is Y = M (Θ), Where θ represents the input parameters, specifically the physical properties; θ=(θ1,θ2,…,θ k ) is a vector containing k input variables, where k is the index of the input variable; The sensitivity analysis of each physical property is performed to evaluate the influence of each physical property on the behavior of the Lamb wave finite element model, and the key parameters in the input parameters are determined to include: The Sobol index is used to perform sensitivity analysis on the output response Y of the Lamb wave finite element model in different dimensions; The Sobol index is analyzed by Monte Carlo simulation to evaluate the influence of each physical property on the behavior of the Lamb wave finite element model and determine the key parameters among the input parameters; The decomposition expression of the Lamb wave finite element model output response in different dimensions is: The constant Μ0 is the mean value of the output response of the Lamb wave finite element model, M i (θ i ) represents the conditional mean of input parameter i, M ij (θ i ,θ j ) represents the conditional mean of input parameter ij when input parameter i and input parameter j (i≠j) interact with each other; M 1…k (θ1…θ k ) represents the conditional mean of input parameters 1…k when k input parameters interact with each other; M i ((θ i ) is expressed as ~i means that the parameter θ is excluded i ; M ij (θ i ,θ j ) is expressed as ~ij means that the parameter θ is excluded ij ; By calculating the first-order Sobol index of input parameters and output responses, the sensitivity measurement results of individual physical properties are obtained; The expression of the first-order Sobol index is Var represents variance; By calculating the second-order Sobol index of input parameters and output responses, we can obtain the sensitivity measurement results when two physical properties interact with each other. The expression of the second-order Sobol index is 7. The method for calibrating physical property parameters of a composite material plate according to claim 6, characterized in that: The agent model uses Bayesian inference and Monte Carlo simulation to randomly update physical property parameters, including: Determine the prior distribution π(Θ) of key parameters among the input parameters based on a deterministic update process; Determine the posterior distribution density function of the input parameter Θ according to the prior distribution π(Θ); The calculation formula of the posterior distribution density function is: in, is the observed random data, is the conditional likelihood function, is the marginal likelihood function; The Monte Carlo sampling algorithm is used to sample the posterior distribution density function to generate a series of random variables; Based on the generated random variables, simulation experiments are carried out to obtain experimental results, and parameters are estimated according to the experimental results to obtain the predicted values of the physical property parameters of the current composite material plate.
8. A physical property parameter calibration system for a composite material plate, used to implement the physical property parameter calibration method for a composite material plate according to any one of claims 1 to 7, characterized in that: include: An acquisition module, used to acquire initial information of input parameters of the composite material plate, wherein the input parameters include a number of physical property parameters capable of characterizing the mechanical properties of the composite material plate; A simulation module is used to input the initial information of the input parameters into the Lamb wave finite element model for simulation to obtain a simulation signal; An acquisition module is used to acquire Lamb wave signals on the composite material plate in real time as experimental signals; A deterministic updating module is used to deterministically update the initial information of the input parameters using simulation signals and experimental signals to obtain the optimal property value of each physical property parameter in the input parameters; The random update module is used to input the optimal property value of each physical property parameter into the proxy model. The proxy model uses Bayesian inference and Monte Carlo simulation to iteratively and randomly update the physical property parameters to obtain the predicted values of the physical property parameters of the current composite material plate.
9. An electronic device, characterized in that: The electronic device comprises: at least one processor (10); and, a memory (11) communicatively connected to the at least one processor (10); The memory (11) stores a computer program executable by the at least one processor (10), and the computer program is executed by the at least one processor (10) so that the at least one processor (10) can perform the method for calibrating physical property parameters of a composite material plate as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer storage medium stores a computer program; when the computer program is executed by a processor, the method for calibrating physical property parameters of a composite material plate according to any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Energy storage system simulation method and device based on proxy model, equipment and medium
CN120297169A
Energy storage system simulation method and device based on agent model, equipment and medium
CN120297169B
