Prediction of the mixed mode I/II fracture toughness of PMMA by an artificial intelligence approach

Artificial intelligence is playing an increasing role in materials testing, whether it is in a new material design, designing new testing methods, or creating a model to predict materials properties. In this research, the artificial intelligence was from an artificial neural network (ANN) and an adaptive neuro-fuzzy inference system (ANFIS), which was applied to predict mixed mode I/II fracture toughness of polymethyl methacrylate material (PMMA). The predictive modeling was based on the factors of thickness, width, crack length to width ratio of the specimen, and mode mixity angle. The training, validation, and testing process of the model used a total of 96 data points per factor. The efficiency of the ANN model in the modeling process, R-2, MSE and MAPE, was 0.9905, 0.0859, and 4.7911 for mode I fracture toughness and 0.9848, 0.0161 and 4.1994 for mode II fracture toughness. The efficiency of the ANFIS model in the modeling process, R-2, MSE and MAPE, for mode I fracture toughness was 0.9953, 0.0415, and 3.2601, while for mode II fracture toughness was 0.9894, 0.0112, and 3.0894. The model application is used to predict the fracture toughness at different levels of factors from the modeling process, with results showing that the fracture toughness from the prediction model is slightly different from the experimental values.

Automated summary

Artificial intelligence is increasingly used in materials testing for tasks such as new material design and predicting materials properties. In this particular study, artificial neural networks and adaptive neuro-fuzzy inference systems were used to predict fracture toughness of PMMA. The models showed high accuracy in predicting fracture toughness under different conditions, although there were slight discrepancies when compared to experimental values.