A new semi-analytical method, discussed in the presented paper, is composed of two stages. Stage A corresponds to the direct analysis, in which the Lamb Waves Measurements (LWM) technique enables obtaining an experimental set of points , where f and k are frequency and wavenumber, respectively. After the preprocessing in the transform space an experimental approximate curve can be formulated. In Stage B the identification procedure is simulated as a sequence of direct analyses. The dimensionless Lamb Dispersion curves are computed by means of the dimensionless simulation curve k_sim ( f | par), where the vector of plate parameters par = {E, v, d, p} is adopted, in which Young's modulus E , Poisson ratio v , plate thickness d and density p are used. The main idea of the proposed approach is similar to that in the classical method of error minimization. In our paper we propose to apply the zero error value of relative criterion Reky = 0, cf. formula (15). The formula can be applied for the identification of a single plate parameter, assuming a fixed value of the other plate parameters. This approach was used in a case study, in which Stages A and B were analysed for an aluminum plate.


Keywords: Structure Health Monitoring, non-destructive method, Lamb waves, dispersion curve, modes of vibration, elastic isotropic and homogenous plate, identification of plate parameters.

Sequential stochastic identification of elastic parameters of thin aluminum plates using Lamb waves is proposed. The identification process is formulated as a Bayesian state estimation problem in which the elastic constants are the unknown state variables. The comparison of a sequence of numerical and pseudo-experimental fundamental dispersion curves is used for an inverse analysis based on particle filter to obtain sequentially the elastic constants. The proposed identification procedure is illustrated by numerical experiments in which the elastic parameters of an aluminum thin plate are estimated. The results show that the proposed approach is able to identify the unknown elastic constants sequentially and that this approach can be also useful for the quantification of uncertainty with respect to the identified parameters.


Keywords: Bayesian state estimation, particle filter, guided Lamb waves, dispersion curves, thin plate.

In this paper an evolutionary algorithms (EA) application to the physically based approximation (PBA) of experimental and/or numerical data is considered. Such an approximation may simultaneously use the whole experimental, theoretical and heuristic knowledge about the analyzed problems. The PBA may be also applied for smoothing discrete data obtained from any rough numerical solution of the boundary value problem, and for solving inverse problems as well, like reconstruction of residual stresses based on experimental data. The PBA presents a very general approach formulated as a large non-linear constrained optimization problem. Its solution is usually complex and troublesome, especially in the case of non-convex problems. Here, considered is a solution approach of such problems based on the EA. However, the standard EA are rather slow methods, especially in the final stage of optimization process. In order to increase their solution efficiency, several acceleration techniques were introduced. Various benchmark problems were analyzed using the improved EA. The intended application of this research is reconstruction of residual stresses in railroads rails and vehicle wheels based on neutronography measurements.


Keywords: evolutionary algorithms, solution efficiency increase, experimental data smoothing, large non-linear constrained optimization problems.

In this paper we present a new approach for solving identification problems based on a novel combination of computer vision techniques, Bayesian state estimation and finite element method. Using our approach we solved two identification problems for a laboratory-scale aluminum frame. In the first problem, we recursively estimated the elastic modulus of the frame material. In the second problem, for the known elastic constant, we identified sequentially the position of a quasi-static concentrated load.


Keywords: identification problems, Bayesian state estimation, particle filtering, computer vision, digital image correlation, finite element method.

The paper presents the application of artificial neural networks (ANN) for description of the Ramberg- Osgood (RO) material model, representing the non linear strain-stress relationship of ε (σ). A neural model of material (NMM) is a feed-forward layered neural network (FLNN) whose parameters were determined using the penalized least squares (PLS) method. A FLNN performing the inverse problem: σ(ε), using pseudo empirical patterns, was developed. Two models of NMM were developed, i.e. a standard model (SNN) and a model based on Bayesian inference (BNN). The properties of the models were compared on the example of a reference truss structure. The computations were performed by means of the hybrid FEM/NMM program, in which NMM developed previously described the current model of the material, and made it possible to explicitly build a tangent operator E_t = dσ/dε. The neural model of material was applied to the analysis of the shakedown of load carrying capacity of an aluminum truss.


Keywords: artificial neural network, inverse problem, material modeling, finite element method, hybrid program, shakedown analysis.

Two problems are presented in the paper concerning axial loading of R/C columns: I) prediction of critical loads, II) identification of concrete strength. The problems were analyzed by two methods: A) Gaussian Processes Method, B) Advanced Back-Propagation Neural Network. The results of the numerical analysis are discussed with respect to numerical efficiency of the applied methods.


Keywords: Gauss Processes Method (GPM), Advanced Back-Propagation Neural Network (ABPNN), Reinforced Concrete (R/C), axial loading, Success Ratio (SR).

This paper presents application of artificial neural networks (ANNs) for prediction of consistency parameters (plastic limit, liquid limit) of fen soils in comparison with the standard regression analysis. All samples of cohesive soils were retrieved from the Wisłok river floodplain, in the vicinity of Rzeszów, near Lisia Góra (Fox Mountain) reserve. Basic fractions (clay, silt, sand) of fen soils are independent variables in modeling of soil properties. Two regression models and a standard multi-layer back-propagation net have been used.


Keywords: fen soils, granulation, plastic limit, liquid limit, regression, artificial neural networks.

The article presents basic rules for constructing and training neural networks, called the Support Vector Machine technique. SVM networks can mainly be used for solving tasks of classification of linearly and nonlinearly separable data and regression as well as identifying signals and recognising increases. In this paper SVM networks have been used for classifying linearly separable data in order to formulate a model of displacements of points representing a monitored object. The problem of learning networks requires the use of quadratic programming in search of an optimum point of a Lagrange function with respect to optimised parameters. Estimated parameters determine the location of the hyperplane which maximises the separation margin of both classes.


Keywords: linear SVM network, classification, displacements.