Structural state assessment based on similarity degree of membership cloud

Most methods of structural states assessment focused on statistical analysis considering only random uncertainties, or on fuzzy recognition considering only fuzzy uncertainties. The cloud model considers both of the two, implementing the transform between qualitative evaluation and quantitative numerical value. In this paper, healthy observations are used to determine the expectation function and feature parameters of the cloud model of healthy states. With the prescribed similarity matrix of structural states, an optimal group of expectation functions and feature parameters representing the other damage states can be inferred as an inverse problem from healthy observations only. On the basis of above, the membership cloud of the four standard damage states can be established using the cloud generator, so could any unknown state. Similar degree between the membership clouds of the unknown state and the standard damage states can be calculated. Lastly, the unknown state is assessed to one of the standard states based on the maximum similar degree. Case study validates the effectiveness and practicability of the above method. This method could be embedded into the Structural Health Monitoring (SHM) system to show a quantitative specification of damage degree and work for the structural maintenance.