Simultaneous identification of structural damage and nonlinear hysteresis parameters by an evolutionary algorithm-based artificial neural network

A new method based on dynamic response sensitivity is proposed for the identification of local nonlinear system. The unknown local nonlinear hysteresis parameters are identified together with linear structural parameters. The sensitivities of structural dynamic response with respect to unknown parameters are derived. The identification equation is set up based on Taylor’s first order approximation, and is solved with the damped least-squares method. A four-story shearing system with rubber isolation bearing of nonlinear hysteresis model is studied to validate the proposed method. Numerical simulation with noisy measured accelerations shows that the proposed method can accurately identify both linear and nonlinear parameters of the system. This method provides a new approach for detecting local nonlinear structures with incomplete measured output information.

Triplet shear tests are commonly used to evaluate the strength of masonry shear walls. A conventional approach involves identifying the parameters of the Mohr-Coulomb failure criterion. Although numerous studies have proposed numerical models capable of reproducing the global experimental response, the present work distinguishes itself by coupling full-field displacement measurements obtained through Digital Image Correlation (DIC) with the identification of nonlinear material parameters within a consistent inverse framework. DIC measurements were used to prescribe experimentally measured boundary conditions in finite element simulations. A Finite Element Model Updating (FEMU) strategy was then implemented to identify the model parameters. A prior sensitivity analysis showed that displacement fields exhibited limited influence on the targeted parameters for the adopted constitutive model, which justified a force-based identification procedure. The methodology was applied to four triplet shear tests. The results demonstrated the capability of the mesoscopic model to reproduce the observed mortar joint response. Both elastic and nonlinear parameters were identified, and their associated uncertainties were quantified.

This paper aims at applying the adaptive Tikhonov regularization to realize the simultaneous identification of structural physical parameters and input based on dynamic response sensitivity. The theories of the simultaneous identification algorithm and the adaptive Tikhonov regularization were first briefly introduced, and then a twelve story shear frame was studied for validating the proposed method. Numerical simulations with noisy measured accelerations show that the proposed method can accurately identify structural parameters and unknown support excitation from only several responses of the structure, and this method can provide a new approach for structural model updating or damage detection with unknown input and incomplete measured output information.

A structural parametric identification strategy based on neural networks algorithms using dynamic macro-strain measurements in time domain from a long-gage strain sensor by fiber optic sensing technique such as Fiber Bragg Grating (FBG) sensor is developed. An array of long-gage sensors is bounded on the structure to measure reliably and accurately macro-strains. By the proposed methodology, the structural parameter of stiffness can be identified. A beam model with known mass distribution is considered as an object structure. Without any eigenvalue analysis or optimization computation, the structural parameter of stiffness can be identified. First an emulator neural network is presented to identify the beam structure in current state. Free vibration macro-strain responses of the beam structure are used to train the emulator neural network. The trained emulator neural network can be used to forecast the free vibration macro-strain response of the beam structure with enough precision and decide the difference between the free vibration macro-strain responses of other assumed structure with different structural parameters and those of the original beam structure. The root mean square (RMS) error vector is presented to evaluate the difference. Subsequently, corresponding to each assumed structure with different structural parameters, the RMS error vector can be calculated. By using the training data set composed of the structural parameters and RMS error vector, a parametric evaluation neural network is trained. A beam structure is considered as an existing structure, based on the trained parametric evaluation neural network, the stiffness of the beam structure can be forecast. It is shown that the parametric identification strategy using macro-strain measurement from long-gage sensors has the potential of being a practical tool for a health monitoring methodology applied to civil engineering structures.

In order to improve the radiation-resistance ability of the InP based high electron mobility transistor (InP HEMT) by optimizing the epitaxial structure design, a series of InP HEMT epitaxial structure materials with different structure parameters is grown by gas source molecular beam epitaxy. These samples are irradiated at room temperature by a 1.5-MeV electron beam at the same irradiation fluence of 2 × 10<sup>15</sup> cm<sup>–2</sup>. The electrical properties of the two-dimensional electron gas (2DEG) for InP HEMT epitaxial materials before and after irradiation are measured by Hall measurements to obtain the changes of the normalized 2DEG density and electron mobility along with the epitaxial structure parameters. The relation between 2DEG radiation damage and epitaxial structure parameters (such as Si-δ-doping density, spacer thickness, channel thickness and channel In content) of InP HEMT epitaxial structure materials is analyzed. The results show that the 2DEG of the InP HEMT epitaxial structure material with higher Si-δ-doping density, thinner spacer thickness, thicker channel thickness and lower channel In content has lower radiation damage, which possesses the stronger radiation-resistance ability.

In principal, a proper analysis of the dynamic response of a structure can provide general indicators of its operational conditions. When the dynamic response changes due to a variation of the physical properties of a structure, then one may conclude that some kind of damages has occurred. This paper presents investigation of the robustness and comparison of four simple methodologies to both identify and quantify the damages in structures, based on the use of Frequency Response Functions (FRF) signals, Principal Component Analysis technique (PCA) and Transmissibility. A steel beam with constant rectangular cross-section is used to compare the proposed approaches. At first, nine damaged scenarios are created and for each of them numerical examples are discussed; a database of FRFs is measured using modal testing. Then, PCA theory is applied to the FRF matrix and global damage detection and quantification indices are defined by using the first 3 Principal Components; Hotelling's T-squared distribution is also applied and by using transmissibility two other indicators, Transmissibility Damage Indicator and Weighted Damage Indicator, are computed for the assessment of damage. The reported examples show that all proposed methods are able to detect and quantify damages at the initial stage.

Two-stage structural damage detection using fuzzy neural networks and data fusion techniques

Progressive damage evolution in rock masses serves as the fundamental mechanism driving geological hazards by controlling deformation patterns and failure predictability. To address the critical challenge of predicting fracture behaviors in heterogeneous geological media, this study pioneers the integration of real-time computed tomography (CT) scanning and acoustic emission (AE) monitoring to investigate self-organized criticality and fracture predictability in Cretaceous sandstone under uniaxial compression. By systematically analyzing internal structural evolution and damage parameters, this established a multiparameter framework to characterize self-organized processes and critical phase transitions during progressive fracturing. Key findings include the following: (1) Distinct critical thresholds emerge during yield-stage self-organization, marked by abrupt transitions in AE signals and crack metrics—from microdamage coalescence initiating volumetric expansion (first critical point) to macrocrack nucleation preceding peak strength (second critical point). (2) AE-crack evolution follows power–law statistics, where elevated scaling exponents (r > 0.85) correlate with intensified nonlinear damage, accelerated localization, and progressive rate enhancement. Yield-stage power–law acceleration provides quantifiable failure precursors. (3) Yield-stage damage patterns exhibit 85% similarity with terminal failure configurations, confirming yield-stage as the definitive precursor with critical temporal signatures for failure prediction. A conceptual framework integrating multiparameter responses (AE signals, crack metrics) was developed to decipher self-organized critical phase transitions during deformation-failure processes. This work establishes methodological foundations for investigating damage mechanisms and predictive strategies in heterogeneous rock systems.

Correlations between structural damage and ground motion parameters during the Ms8.0 Wenchuan Earthquake

Three sets of real earthquake records are selected to represent seismic ground motions in the low, normal, and high \Ia\N/\Iv\N (peak ground acceleration/peak ground velocity) ranges. The inelastic responses of single-degree-of-freedom stiffness degrading systems to the three sets of records are analyzed statistically to investigate the significance of such ground motion characteristics on structural damage. Both damage due to peak inelastic deformation and cumulative fatigue-type damage due to a large number of reversed inelastic deformations are considered. Three damage parameters, each reflecting information on peak inelastic deformation, stiffness deterioration, and hysteretic energy dissipation, respectively, are used to measure structural damage. The mean values of the three damage indicators are obtained for the three sets of accelerograms, and the corresponding dispersion characteristics are examined. It is found that ground motion \Ia\N/\Iv\N range has a significant effect not only on peak inelastic response but also on hysteretic energy dissipation and stiffness deterioration of stiffness degrading systems. The effect on hysteretic energy dissipation is more pronounced than that of peak inelastic response.

Transcendental inverse eigenvalue problems in damage parameter estimation

In order to realize the quantitative analysis of the structural integrity and damage evolution of hydroxyl-terminated polybutadiene (HTPB) composite solid propellant under external load, the loading failure experiments were carried out on the macro rectangular specimen and the micro specimen, and the numerical analysis of the strain evolution process in the region of interest (ROI) of the specimen was carried out using digital image correlation (DIC). At the same time, with the help of finite element model updating (FEMU) method, the mechanical and cohesion parameters of the specimen were inversed using the combination of adaptive particle swarm optimization (APSO), back propagation (BP) and Hooke-Jeeves algorithm, and the simulation of the whole process of propellant damage and fracture was realized from the microscopic point of view. The results show that inhibiting the debonding of propellant/liner interface is the key to maintain the structural integrity. It starts with the damage and fracture of the propellant side. A scanning electron microscope (SEM) in-situ dynamic tensile test shows that the initial damage occurs at the strain of 27.368%, and the through-type crack propagates along the interface when the strain reaches 43.276%. In addition, the use of combinatorial optimization algorithm can realize the global optimal inversion of 16 parameters divided into three types in 100 complete calculations, reduce the optimal objective function value to 0.0251, and assist the finite element calculation to realize the quantitative analysis and accurate simulation of the experimental process.

Closed cell polymeric foams, for example EVA foams, are widely used in sport and medical equipments such as helmets, seating, footwear and therapeutic plug/insoles. The mechanical behavior of foams is highly nonlinear, and the materials are often under complex loading conditions in service over different spectrums of strain. Foam testing and determination of the nonlinear foam constitutive parameters are important to predict materials performances in service and to aid product design and developments. One promising approach is to use the indentation method combined with inverse finite element (FE) programs. This approach has been extensively studied and several programs such as interactive, parametric and artificial neural networks (ANN) have found successful applications for different conditions. ANN represents a simple and quick method without repeated running of FE program, this could provide a significant advantage. In this study, an ANN based inverse FE program has been developed and used to predict the nonlinear material properties of EVA foams. A 2-D parametric FE model was developed and validated against experimental data. Systematic data from FE simulations was used to train and validate the ANN model. The accuracy and validity of the ANN method were assessed based on both blind tests and experimental data. Results showed that the proposed artificial neural network model is robust and efficient in predicating the nonlinear parameters of foam materials. Finally, potential applications of the method and future improvement are discussed.KeywordsIndentation testEVA foamFinite Element ModellingArtificial Neural Network

Focused ion beam (FIB) techniques have previously been shown to have applications in templating semiconductor nanostructure growth on Si. To assess crystalline quality in this method, or in other FIB-based nano-fabrication methods in Si, we assess in this work FIB-implantation damage in Si and subsequent recovery by annealing. Specifically, we study Si substrates implanted to 1 × 1012–5 × 1015 ions cm−2 fluences of 30 kV Si2+, Ge2+ and Ga+ incident normally at room temperature, covering a structural damage regime from almost no damage to amorphization. Raman spectroscopy with incident photons of wavelength 405 and 514 nm were used for probing structural damage in different depths in as-implanted and thermally annealed substrates. Annealing was performed for varying times at 730–900 °C in ultra-high purity nitrogen ambient. Structural damage quantification was performed through measurements of peak height and position of the crystalline Si peak at 520 cm−1. Our analysis shows that a structural damage parameter, D, (defined by subtracting from unity, the ratio of the height of the Si Raman peak in implanted/annealed samples to that for the Si peak in an unimplanted standard, such that D = 0.00 and D = 1.00 correspond to pristine single crystal and amorphous Si respectively) is 0.05 or less in all the Si and Ge implants after annealing to 900 °C, 1800 s. However, D for the Ga implants is close to zero for Ga fluences up to 3 × 1013 cm−2 but increases steadily to 0.19 for 5 × 1015 Ga cm−2 under these annealing conditions. The underlying damage recovery mechanisms are discussed.

The method of evaluating the level of residual surface stresses in the products without the loss of their integrity at elastoplastic strains and under static and cyclic loading by scattering of hardness characteristics is proposed. Residual surface stress levels are assessed not by absolute hardness values but by the state of material structure. Its homogeneity coefficient in the Weibull distribution, characterizing the scattering of large-scale hardness data, was taken as a calculated structure damage parameter and evaluated by the LM-hardness method based on direct multiple independent hardness measurements. The estimation of residual stresses is built upon the correlation relation between the material structure damage and residual stresses. By comparing the measured parameters of structure homogeneity, charactering their damage level in the region of residual stresses, with the structure parameter in the region free from them, their level is determined. In the absence of such a region, residual stresses are evaluated by comparing the homogeneity parameter of the product element in their region with that of a reference sample made of the same material and subject to the same operating loads as the examined product. The product quality can be estimated by the regions of high residual stress concentrations, comparing the parameters obtained in its different characteristic sections. Correlation relations, determining the values of residual stresses, are found from preliminary experimental tests of the material specimens by constructing the structural parameter – stress diagram at a preset loading step. The structure parameter is measured by scattering its large-scale hardness measurements.