Mechanistic-Based Characterization of Non-Linear Pavement Mechanical Properties with Evolving Intelligent Information Processing Systems

The backcalculation methodology is an inverse process to determine in-situ materials stiffness of pavement layer from Non-Destructive Test (NDT) surface deflections using iterative optimization or more recently, computational intelligence techniques. Intelligent Information Systems (IIS) have been developed with the idea of dealing with information in a way a human expert would by incorporating abilities to learn fast from a large amount of data, dynamically create new modules, connections and neurons, accommodate imprecise or uncertain knowledge, etc. Evolving Connectionist Systems (ECOS) are IIS created with such requirements in mind. This paper explores the feasibility of applying an ECOS methodology, namely Dynamic Evolving Neuro-Fuzzy Inference System (DENFIS), to backcalculation of non-linear pavement mechanical properties and real-time non-destructive condition evaluation of flexible pavements.