Assessing Durability of Structures Exposed to Stationary and Nonstationary Random Loads Using Variation Modal Decomposition

In the current era of digital transformation, most aggregates are equipped with various sensors providing information for the development of smart models to forecast durability. When developing digital twins of natural objects to assess the remaining lifetime, the most likely issues to be addressed deal with algorithmic support or knoware. These problems start manifesting themselves in the adaptive and automatic analysis of signals generated by real sensors installed on the object, and subsequent processing to assess the object durability. We suggest a new approach to assess the structural durability based on applying the adaptive method of variation modal decomposition sensor signals including the nonstationary signals. Variational mode decomposition involves breaking down the initial complex random process into simpler random processes (modes). These are narrowband stationary modes. We apply the initial technique to sum damages sustained based on effects of modal decomposition. The method can be used as a knoware in developing digital twins to predict the remaining lifetime of physical objects. A comparison of the proposed method with the rainflow technique in the time domain is made. Different implementations of random processes are generated to proceed with this comparison including stationary and nonstationary processes having different bandwidth. The comparison implies calculating the synthesized random processes. In addition, the traditional Dirlik and Benasciutti frequency methods are compared with the rainflow method. The proposed method gives good results as the error is less than that provided by traditional frequency techniques, especially for nonstationary processes.