Piotr Nazarko

Soft computing methods in the analysis of elastic wave signals and damage identification

Abstract An outline of a Structure Health Monitoring (SHM) system is given. A non-destructive technique of elastic wave propagation was used in the presented approach. For the analysis of time signals recorded by piezoelectric transducers, signal processing algorithm has been developed and artificial neural networks have been used. As a consequence, two levels of the damage identification problem have been realized: novelty detection and damage assessment. The systems accuracy and reliability have been verified during laboratory tests of strip specimens made of various materials. It has been proved that the system can be used for the analysis of simple as well as complex signals. Moreover, the system can operate online as an automatic SHM system.

Towards application of soft computing in structural health monitoring

The paper presents preliminary results of data analysis and discusses the application of soft computing methods in the field of non-destructive tests. The main objective of developed diagnostic system are the automatic detection and evaluation of damage. Thus the system is composed of two signal processing techniques known as novelty detection and pattern recognition. For this purpose autoassociative as well as feed-forward neural networks are used. All the signals used for training the system are obtained from laboratory tests of strip specimens, where phenomenon of elastic wave propagation in solids was utilized. Computed parameters of time signals defines various types of input vectors used for training neural networks. The results finally obtained prove that the proposed diagnostic system made automation of structure testing possible and can be applied to Structural Health Monitoring.

Application of Eddy Current Sensor System and LDV Device for Ultrasonic Vibrations Measurements

The article presents eddy current sensor system in respect to the measurements of vibrations for ultrasonic assisted machining processes. The comparison of eddy current sensor system with laser doppler vibrometer (LDV device) is also presented. The paper concerns the analysis of the influence of the distance between the tool and the sensor on amplitude value, filtering of raw voltage signals in MATLAB and amplitude values presentation for different tools after MATLAB and LDV software analysis. Functionality of two applied devices can be discussed after investigations. Differences in recorded amplitude’s values are also discussed. The examples of measurements and test stand configuration are presented. For better understanding of hybrid machining process, it is important to indicate the possible methods of amplitude measurements and to analyze the results of measurements.

Analysis of Lamb wave dispersion curve sensitivity to material elastic constants in composites

The paper deals with the problem of Lamb waves dispersion curves sensitivity to the change of elastic constants in composite materials. The framework of the present work is a more general problem of material constants identification in thin plates made of composite materials. The approach is based on the analysis of guided waves propagation and the related dispersion curves to find the underlying material elastic constants. In present work a numerical study is performed to identify measurement directions and wave propagation modes that are most sensitive to the change of the particular elastic constants. This approach will allow to optimize the material constants identification procedure and experimental setup by specifying the preferred measurement directions and wave propagation modes. The approach can be used within the Structural Health Monitoring framework to monitor material degradation of plate-like structures made of composite materials.

DEVELOPMENT OF A HYBRID SYSTEM FOR IDENTIFICATION OF ELASTIC MATERIAL IN COMPOSITE LAMINA: APPLICATION OF LAMB WAVES AND NEURAL NETWORKS

A new Hybrid Computational System is presented, developed for the identification of homogeneous, elastic, hexagonally orthotropic plate parameters. Attention is focused on the construction of dispersion curves, related to Lamb waves. The main idea of the hybrid system lies in the separation of two essential basic computational stages, corresponding to the direct and inverse analyses. In the frame of the first stage an experimental dispersion curve DCexp is constructed, applying Guided Wave Measurement technique. The other stage is supported on the application of artificial neural network trained ‘off line’. Two case studies are presented, corresponding either to pseudo-experimental computer simulations or to laboratory tests.

Experiments of Damage Detection in Strips Based on Soft Computing Methods andWave Propagation

All industry branches like aerospace, mechanical and civil engineering are interested in less intrusion and more accuracy failure assessment techniques. They are mostly interested in damages like cracks, delaminations, disbanding, corrosion, etc. Damage detection and assessment technique was developed in this paper. It uses variations in structural wave propagation for undamaged and damaged structure. This Structural Health Monitoring (SHM) method is useful especially in large, complex and inaccessible structures [1], [2]. Based on earlier promising results with this approach [3], [4] a set of laboratory tests were carried out on simple elements like strips. Two kind of materials were used: steel and plexy. Several failure cases were introduced by cutting or drilling the samples. Piezoceramics (PZT) elements were served as transmitters and receivers of elastic waves trough the monitored specimens. During these experiment different groups of excitation signals (continuous sine wave, one, four and six sine wave impulses) and frequency (frequency range from 2 to 50 kHz) were applied to introduce wave to the structure. The numerical models were also created using Finite Element Method (FEM). Defects in the form of a notch were simulated by the removal of selected finite elements from the model. This simulation gave possibility to extend set of damages cases and improved nets generalization properties. In both laboratory and numerical experiments advanced signal processing techniques were adopted. The measured signals were preprocessed by wavelet transform in order to remove noise. Frequency analysis was carried out by Fast Fourier transform (FFT). Replication technique was adopted to experimental data. To realize dependences between input (harmonic frequencies) and output data (height, width and localization of damage) Artificial Neural Networks (ANNs) were used. Several input combinations and nets architectures were tested. Results presented in this paper proved reliability and usefulness of proposed approach.