Structural Damage Detection With Sensor Degradation Into Consideration

It is the key and core of structural health monitoring to effectively collect structural responses and used them for damage detection in civil engineering.In recent years,the research and application on structural damage detection based on vibration have been received increasing attention from researchers and engineers.However,in present damage detection methods,only the effect of loading and environment on damage is considered for the measured responses.Due to the fact that the sensor is deteriorated and is even in fault with the variation of time and environment,it seriously influences the diagnosis ratio and increases the detection difficulty.In addition,the sensor degradation often leads to the incomplete data problems,which are also the hot spot and difficult problems in structural damage detection and structural assessment.As a result,considering the sensor degradation,this thesis studied on structural novelty detection,the degraded sensors isolation,data reconstruction from degraded sensors and structural damage detection and assessment.The main works and achievements are as follows:(1)This thesis proposed a structural novelty detection method with sensor degradation into consideration.Firstly,considering the effects of measurement accuracy and environmental variations on measurement variance,the author proposed an adaptive consensus data fusion algorithm,which is able to choose sensors whose data will be subsequently fused.Secondly,due to the fact that discrete wavelet transforms(DWT)influences by noise,the author presented a DWT-RouboustICA method to accurately detect structural novelty.Finally,simulation of the seven-story frame was carried out to demonstrate that the proposed method can detect structural damage novelty and sensor degradation novelty,and which has excellent performance in terms of robustness and anti-noise.Furthermore,when the noise level is 10%,the proposed method also can identify structural novelty.(2)This thesis presented a method to locate the novelty of structural responses and distinguish from the sensor degradation and structural damage.Therein,the root mean square of the generalized likelihood ratio test was firstly used to detect and localize the novelty of structural responses;in the second stage,a new index was proposed and then performed in the statistical process control chart to distinguish from the sensor degradation and structural damage.Herein,the new index was the percentage of the extreme value of the largest principal component scores of the generalized likelihood ratio.The efficiency and applicability of the proposed approach were validated by numerical simulations of a planar truss structure and a simply-supported steel beam experiment in laboratory.The results show that:the proposed method can effectively locate damage with the sensor degradation and environmental impacts into consideration.In addition,compared with other index,the new index was able to accurately and quickly judge the source of the change in responses by statistical process control chart.(3)This thesis proposed a data recovery method based on improved multi-scale principal component analysis(IMSPCA)and adaptive consensus data replacement methods.More specifically,as for data collected from degraded sensors,the author utilized cross correlation coefficient in discrete wavelet transform(DWT)to judge the location of storing effective structural responses in order to replace this coefficient and then the MSPCA algorithm is used to data reconstruction.Besides,as for fault sensors,the author proposed an adaptive consensus data construction method where the data from adjacent sensors were used to replace the incomplete data.The efficiency and applicability of the proposed approach were validated by the numerical simulation of a spatial truss and data collected from the real structural health monitoring system installed in Tianjin Yonghe Bridge.The result provided relativity reliable data for further structural damage detection.(4)Due to the fact that the multi-particle swarm co-evolution optimization(MPSCO)is likely to end up with a premature convergence and often takes a long time to search for the excellent local optimum with in the region of convergence,the author proposed a two-stage damage detection method based on modal stain energy change ratio(MSECR)and improved MPSCO(IMPSCO).In the first stage,the modal strain energy change ratio was used to roughly identify the locations of damaged elements via an appropriate MSECR threshold determined by parameter estimation.In the second stage,IMPSCO was utilized to precisely locate and quantify the damage.Finally,numerical simulation of a seven-story frame and dynamic experiment of an spatial truss experiment in laboratory conditions were performed to validate the proposed method,and a comparison was made between the proposed approach and existing methods.The results show that the proposed method can not only effectively locate damage but also accurately evaluate the extent of damage.Meanwhile,it also enjoys good noise-tolerance and adaptability.In addition,the identified error ratio is less than 4.5%.Compared with MPSCO,the proposed method can not only save time but also has excellent noise-tolerance and robustness.(5)This thesis developed a structural damage identification system with sensor degradation into consideration.More specifically,the author detected the structural novelty and isolated degraded sensor firstly;furthermore,the structural responses from degraded sensors were recovered;the reconstructed responses were used to precisely identify structural damage finally.In addition,a seven-story frame structure was performed in order to validate the effectiveness of the proposed method.The results show that 1)the presented method can not only detect the abnormal instant but also isolate the degraded sensors;2)the proposed method is able to recover the structural responses from degraded sensor;3)Finally,the damage detection results has excellent accuracy at last.(6)Finally,all works were summarized,and the future research interests and development trends were put forward.Through the above-mentioned investigation,some technical bottleneck problems can be solved in the community of structural health monitoring,such as misclassifying the sensor degradation and damage,the damage detection method for incomplete measured data under the environmental effect and sensor deterioration.This can provide technical supports for the large structural health monitoring systems.