Compressive Sampling For Structural Health Monitoring And Damage Detection Methods Based On Information Fusion

Structural health monitoring (SHM) has become the hot topics in the world. Generally, a substantial number of sensors are required in SHM systems because of the large scale and complexity of civil structures. Data compression has, therefore, a potentially important role in SHM to keep the system running with high efficiency. Also, data compression is a key investigation project in wireless sensors and networks. Besides, structural damage detection is a key problem of SHM. However, the uncertainties seriously impede reliable identification of damage. Therefore, this thesis mainly focuses on investigating the compressive sampling techniques for SHM and developing the new damage detection methods with considering uncertainties.Main contents are included as follows:The vibration data compression method based on CS is proposed. Firstly, the CS theory is introduced. Then the potential of compressive sampling (CS) for data compression of vibration data is investigated. The relation of CS compression ratio and reconstruction error of vibration data is analysed. The acceleration data collected from the SHM system of Shandong Binzhou Yellow River Highway Bridge and National Aquatics Center are used to illustrate the data compression ability of CS. For comparison, other data compression methods are also employed to compress the data.A new CS method based Bayesian theory is presented. Firstly, the probability model of the wavelet coefficients of signal is constructed using sparse prior probability density function according the Bayesian theory. Secondly, the parameters are solved using Bayesian model selection methods. Thirdly, the mean and variance of wavelet coefficients of signal are calculated. Lastly, the original signal can be reconstructed by inverse wavelet transform using the mean of wavelet coefficients of signal. The acceleration data collected from the SHM system of Shandong Binzhou Yellow River Highway Bridge and National Aquatics Center are used to analyse the data compression ability of BCS. For comparsion, the CS is also employed to compress the data.The design of Analog-to-Digital (ADC) based on CS technique is investigated. Firstly, the CS of analog signal is introduced. Secondly, the design of ADC based on CS (CSADC) is proposed and the circuit diagram of CSADC is presented. Lastly, a numerical example is employed to illustrate the data acquisition ability of CSADC.A damage identification method based on the combination of artificial neural network (ANN), Dempster-Shafer (D-S) evidence theory-based information fusion and the Shannon entropy is proposed. The initial damage decision is first made by several individual ANNs with different inputs. Considering that damage identification accuracy is dependent on different ANNs, optimal weighting coefficients obtained by Genetic Algorithm (GA) are assigned to ANNs. Damage identification-based D-S evidence theory is carried out by combining the decision of the ANN with the largest weighting coefficient with that of the ANN with the second largest weighting coefficient. The decision with smallest entropy remains in the next information fusion operation because this decision has less uncertainty. The decision with smaller entropy will be combined with the decision of the ANN with the third largest weighting coefficient. The operation is repeated until the last ANN with the minimum weighting coefficient is fused. Numerical study on the Binzhou Yellow River Highway Bridge and experiment sdudy on a real Bridge was carried out to validate the accuracy of the proposed damage identification method.A D-S evidence theory based approach for on-line structural damage detection is proposed. Firstly, the Bayesian method is employed to calculate the damage probability of substructures using the multi-sensors data sets obtained from the structure, and the damage probabilities of substructures are transformed to damage basic probability assessments which used in evidence theory. Then the D-S evidence theory is employed to combine the individual damage basic probability assessments for getting the last damage detection results. With considering multi-sensors data including acceleration and strain, and measurement noise the numerical studies on a 14-bay planar rigid frame structure are carried out. The results indicate that the damage detection results obtained by combining the damage basic probability assessments from each test data are better than the individual results obtained just by each test data separately. The damage detection results based on Bayesian updating methods are also given to compare the difference between D-S and Bayesian on damage detection.