Fused deposition modeling (FDM) is one of the most popular additive manufacturing technologies for fabricating prototypes with complex geometry and different materials. However, current commercial FDM machines have the limitations in process reliability and product quality. In order to overcome these limitations and increase the levels of machine intelligence and automation, machine conditions need to be monitored more closely as in closedloop control systems. In this study, a new method for in situ monitoring of FDM machine conditions is proposed, where acoustic emission (AE) technique is applied. The proposed method allows for the identification of both normal and abnormal states of the machine conditions. The time-domain features of AE hits are used as the indicators. Support vector machines with the radial basis function kernel are applied for state identification. Experimental results show that this new method can potentially serve as a nonintrusive diagnostic and prognostic tool for FDM machine maintenance and process control.

No abstract

No abstract

No abstract

Training machine learning tools such as neural networks require the availability of sizable data, which can be difficult for engineering and scientific applications where experiments or simulations are expensive. In this work, a novel multi-fidelity physics-constrained neural network is proposed to reduce the required amount of training data, where physical knowledge is applied to constrain neural networks, and multi-fidelity networks are constructed to improve training efficiency. A low-cost low-fidelity physics-constrained neural network is used as the baseline model, whereas a limited amount of data from a high-fidelity physics-constrained neural network is used to train a second neural network to predict the difference between the two models. The proposed framework is demonstrated with two-dimensional heat transfer, phase transition, and dendritic growth problems, which are fundamental in materials modeling. Physics is described by partial differential equations. With the same set of training data, the prediction error of physics-constrained neural network can be one order of magnitude lower than that of the classical artificial neural network without physical constraints. The accuracy of the prediction is comparable to those from direct numerical solutions of equations.