One of the most effective approaches to computational modeling of composite systems is the homogenization method [1]. A necessary condition for applying the homogenization method is the presence of a certain scale relationship between the components of the reinforced composite and the entire system [2]. Most often, two-scale analytical models are introduced, each of which is associated with a pre-set scale parameter. To implement the effective computational processes, this parameter is specified as a small value, namely, a real number (usually tending to zero). The impossibility of introducing more than two different scales for a local volume of a composite and, in addition, the insufficient sensitivity of the homogenized characteristics of the composite to the geometric relationships of scales can be attributed to the significant disadvantages of such methods [3].
