Damage detection of structures under changing environmental and operational conditions using the COVRATIO statistic.

Damage detection of civil engineering structures is usually carried out by analysing the changes in vibration properties of the structures due to the effects of damage. However, these vibration properties are also affected by the environmental and operational conditions the structures face, which can create false alerts. To tackle these effects, it is proposed to consider damage detection as an outlier analysis case study in this paper. The observations obtained from the damaged condition of the structures have abnormal values of measurements (e.g. abnormal values of natural frequencies) and do not follow the pattern of the undamaged measurements obtained under different environmental and operational conditions. These damaged observations are similar to the outlier measurements. Both types of measurements do not follow the pattern of the normal observations. The COVRATIO statistic, which is usually used to identify outlier measurements, is proposed to be used in this paper to detect damage. The identified outlier measurements are then the damaged observations in this paper. A beam structure and the Z24 Bridge in Switzerland are analysed using the proposed method. The traditional regression damage detection method is also applied to both structures, and it is found that better results are obtained using the method proposed in this paper. 10% and 87% more damaged cases are detected in the beam structure and the Z24 Bridge using the proposed method, respectively. • Damage detection is affected by the environmental and operational conditions. • Damage detection is considered as an outlier analysis case study. • A method using the COVRATIO statistic is proposed for damage detection. • The proposed method successfully classified between undamaged and damaged cases. • The proposed method has better performance than the traditional regression method. [ABSTRACT FROM AUTHOR]