Predictive model of the modulus of elasticity in static bending (MOE) of larch wood based on gray relation analysis (GRA) and gene expression programming (GEP)

Abstract

To accurately evaluate the modulus of elasticity in static bending (MOE) of wooden components in ancient timberwork buildings under the “minimum intervention principle,” the nondestructive testing of physical and mechanical properties were conducted on larch. Using moisture content (MC), density (ρ), the stress wave propagation velocity ( ), the modulus of elasticity in dynamic bending (Ed), the rotational resistance value of the drilling needle (fdrill), and the resistance value of the feeding needle (ffeed) as the main parameters, the correlation between several parameters and MOE was firstly calculated using the Gray Relation Analysis (GRA) and ranked according to the strength of the correlation. Six combinations were selected according to the ranking, and the Gene Expression Programming algorithm (GEP) was used to build models for predicting MOE. The results showed that the correlation between several parameters and MOE was good (between 0.5 and 0.8), and the prediction model established with combination 6 was the best, which indicated that the prediction model established based on GRA-GEP algorithm had a certain feasibility and effectiveness, and the combined effect of the six parameters to evaluate the MOE of wooden components of ancient buildings was better in the field inspection.