This research focuses on the investigation of delamination assessment in composite structures by using methods based on structural dynamic responses. Delamination is a common type of damages for composite structures in applications. It is necessary to detect and assess the delamination in composite structures to ensure the composite structures operating and maintaining. However, the dynamic responses of structures with delamination may be difficult to be analyzed due to the complexity of composite structures as the result of the different properties of materials in the composite structures. Therefore, it is necessary to develop effective methods for delamination analysis and assessment, which will be investigated in this research. To address these problems, this research aims to:
1. Develop new methodologies for dynamic analysis of delaminated composite structures to analyze the effect of delamination based on the evaluation of the vibration characteristics of composite structures with different delamination configuration;
2. Develop effective methodologies to detect and assess delamination in composite structures based on the dynamic responses by using the phase space topology analysis to improve the sensitivity and robustness of delamination assessment for composites structures;
To achieve these two aims, the research content and novelty are stated as:
1) Firstly, this study proposes a method based on the Green’s function to develop analytical models of delaminated structures that can be used to investigate the effect generated by delamination on the vibration characteristics. The accuracy of this developed model to describe the vibration characteristics of delaminated beam structures especially under forced excitation is verified by the comparison with other types of models, including the numerical models. The result demonstrates the accuracy and advantages of the proposed analytical model to investigate the effect of delamination on the vibration characteristics of the beam structures under excitation of various frequencies with different delamination configurations, such as size, location, and depth. It should also be noted that the proposed modeling method is demonstrated useful to investigate the vibration characteristics of various measurement locations, which is important for the delamination assessment based on the dynamic responses of structures.
2) Secondly, based on the developed analytical model by using the Green’s function, the investigation has been done for the effect of various measurement locations on the sensitivity to the particular vibration modes with various delamination configurations. Based on this situation, the methodology based on the modal observability (Mn) and the spatial observability (So) is proposed to optimize the structural sensor locations to make the measurement focusing on the vibration modes sensitive to the delamination, which can improve the delamination assessment. The result demonstrates that the proposed methodology is effective to determine the sensor locations which can provide strong signals with sufficient distributions of the particular vibration modes sensitive to the delamination. So the optimization can improve the delamination assessment effectively based on providing sensitive vibration measurement locations.
3) Thirdly, a methodology based on the phase space topology analysis of the dynamic signals measured from the structures is proposed to assess delamination in composite structures. The phase space topology structures are evaluated by using a method named phase space reconstructed (PSR) method based on the dynamic signals measured by the dynamic sensors. A feature named the change of phase space topology (CPST) is used to describe the effect of delamination on the phase space topology structures. The result demonstrated that the phase space topology structures and CPST are sensitive to delamination. The robustness of the proposed feature to the measured noise has also been tested, which shows that the proposed method and feature have sufficient robustness to the measured noise in applications.
This research also improves the methodology to assess delamination based on the phase space topology analysis by incorporating with the wavelet packet decomposition. The wavelet packet method can decompose a dynamic signal into several sub-signals in different frequency ranges, which may contain different local information relevant to the delamination. Then the phase space topology structures and the CPSTs of different sub-signals can be evaluated and investigated to analyze the local information. The phase space topology structures of sub-signals decomposed by the wavelet packet method can describe the change of energy distribution of sub-signals in different frequency ranges generated by the delamination. The possibility of the proposed method is demonstrated by the simulation and experiment and the proposed features will be used in the following work;
4) Based on the previous work, a method by using the artificial neural network (ANN) based on the phase space topology analysis to estimate delamination in structures is proposed tested. The ANN can be used to describe the relationship between the delamination and vibration characteristics to estimate the delamination without mechanism analysis for composite structures. The CPSTs of original signals and sub-signals decomposed by the wavelet packet method are used as input factors for the ANN to assess the delamination in composite structures. The accuracy of the ANN for the delamination assessment can be enhanced by training the ANN with more cases. The possibility and the potential are tested in this research. The different performances for various delamination parameters estimate are also analyzed, which shows that the performance for various delamination parameter assessments is different due to the different effect of delamination on the input factors. Furthermore, the performance of delamination assessment by using the ANN with different input factors is investigated to analyze the effect of input factors on the delamination assessment performance and find the best input factors for ANN in this research. The results show the CPSTs of sub-signals generated by wavelet packet decomposition are the best input factors because this type of feature can provide more information with high sensitivity and good robustness to the measurement noise.
In conclusion, this research will provide a systematic study for the improvement of delamination assessment and development of applicable methods for composite structures based on the dynamic signals by analyzing the phase space topology structures combined with wavelet packet decomposition and ANN. Moreover, the theoretical analysis and optimization for dynamic signal measurement are analyzed to provide explanation and support for the delamination assessment. The potential of the proposed methods for other types of damage in composite structures and other applications are also mentioned in this research.
|Date of Award||8 Jul 2021|
- Univerisity of Nottingham
|Supervisor||Dunant Halim (Supervisor), Chris Rudd (Supervisor) & Xiaoling Liu (Supervisor)|