Bond strength prediction of timber-FRP under standard and acidic/alkaline environmental conditions based on gene expression programming

Timber is widely used as a construction material; however, the environmental deterioration of timber is a crucial problem for the construction industry. Fiber-reinforced polymer (FRP) has been considered appropriate and beneficial for the repair and rehabilitation of timber. This study proposes three empirical models using a supervised machine learning method called gene expression programming (GEP) to predict the bond strength between timber and FRP under various environmental conditions. The first empirical model is used to predict bond strength under standard conditions. The two other models are proposed to predict the strength reduction in acidic and alkali solutions. The formulation variables include duration, pH level, mechanical properties of FRP, types of timber and adhesive, and geometry of the specimens. In this regard, the database consisting of 251 test data is collected from previous studies. Eighty percent of the data is used for training the models, and the rest is applied to validation. Both training and validation data pass statistical criteria for all three models. The mean relative error is smaller than 12%, and the minimum R coefficient is 0.9 besides the appropriated root mean squared error (RMSE) and mean absolute error (MAE) values.