Accelerating the Layup Sequences Design of Composite Laminates via Theory-Guided Machine Learning Models

Abstract

Experimental and numerical investigations are presented for a theory-guided machine learning (ML) model that combines the Hashin failure theory (HFT) and the classical lamination theory (CLT) to optimize and accelerate the design of composite laminates. A finite element simulation with the incorporation of the HFT and CLT were used to generate the training dataset. Instead of directly mapping the relationship between the ply angles of the laminate and its strength and stiffness, a multi-layer interconnected neural network (NN) system was built following the logical sequence of composite theories. With the forward prediction by the NN system and the inverse optimization by genetic algorithm (GA), a benchmark case of designing a composite tube subjected to the combined loads of bending and torsion was studied. The ML models successfully provided the optimal layup sequences and the required fiber modulus according to the preset design targets. Additionally, it shows that the machine learning models, with the guidance of composite theories, realize a faster optimization process and requires less training data than models with direct simple NNs. Such results imply the importance of domain knowledge in helping improve the ML applications in engineering problems.

Neural networks modeling of strain energy density and Tsai-Wu index in laminated composites

“…The prevalent trend is the utilization of laminate configurations as ANN inputs. For example, Liao et al 17 employ two models, each with two ANNs, for strength and stiffness predictions. Liu et al 18 focus on buckling effects.…”

“…where z ðkÞ l correspond to the result of the sum from the product between the weights and the outputs a j lÀ1 of the neurons in the previous layer -size lÀ1 is the number of units in the previous layer, adding the effect of the bias (see equation (17)). In the first layer, the value of a j lÀ1 is replaced by the input vector x.…”

“…As with the previously mentioned articles, there are recent studies where the use of neural networks as metamodels to analyze applications and study the response of laminated composite materials is combined. For example, to predict the strength value of a compound, 17 and for structural buckling analysis, 18 and also for machining optimization in structural applications. 19 In developing metamodels based on ML, the input and output parameters are diverse and focused on the property to be analyzed.…”

Progresses and Challenges of Composite Laminates in Thin-Walled Structures: A Systematic Review

“…To optimize and expedite the design process of composite laminates, Sorrentino 67 presented a theory-guided ML model that combines the Hashin failure theory (HFT) and the classical lamination theory (CLT). The model utilizes a training data set generated through finite element simulations that incorporate the HFT and CLT.…”

Performance analysis of similar and dissimilar self-piercing riveted joints in aluminum alloys

“…10 Liao et al proposed a new joining process called double-sided self-piercing riveting which uses tubular rivets of simple geometry placed between the two parts to be joined. 11 Li et al studied the influence of local friction on the rivet insertion process, joint properties, and static lap shear strength by modifying the local surface of the top sheet around the rivet piercing site with different impression tools. 3 Huang et al proposed an SPR method using a flanged pipe rivet.…”