In organic electronic devices, the interfacial dipole at organic/metal interfaces is critical in determining the carrier injection or extraction that limits the performance of the device. A novel technique to enable the direct measurement of underburied dipoles is developed and demonstrated. By tilting the shadow mask by a small angle, metal atoms diffuse into the opening slit to form an ultrathin metal layer during the evaporation process. As the ultrathin metal layer cannot screen out the dipole-induced surface work function change, the dipole strength and direction at the organic/metal interface can be revealed. It was found that the polarity of the organic material, the Fermi-level pinning and the interface morphology all play important roles in dipole formation. By comparing the energy level shifts at the organic/pre-deposited metal and organic/post-deposited metal interfaces, the dipole formed by molecular interactions could be distinguished from the dipole formed by Fermi-level pinning.

Artificial intelligence technology has enabled the manufacturing industry and actively guided its transformation and promotion for the past few decades. Injection molding technology is a crucial procedure in mechanical engineering and manufacturing due to its adaptability and dimensional stability. An essential step in the injection molding process is quality inspection and manual visual inspection is still used in conventional quality control, but this open-loop working method has issues with subjectivity and real-time monitoring capacity. This paper proposes an integrated “processing–matching–classification–diagnosis” concept based on machine vision and deep learning that allows for efficient and intelligent diagnosis of injection molding in complex scenarios. Based on eight categories of failure images of plastic components, this paper summarizes the theoretical method of processing fault categorization and identifies the various causes of defects from injection machines and molds. A template matching mechanism based on a new concept—arbitration function $J\left({\psi }_{ij}\right)$—provided in this paper, matches the edge features to achieve the initial classification of plastic components images. A conventional VGG16 network is innovatively upgraded in this work in order to further classify the unqualified plastic components. The classification accuracy of this improved VGG16 reaches 96.67%, which is better than the 53.33% of the traditional network. The accuracy, responsiveness, and resilience of the quality inspection are all improved in this paper. This work enhances production safety while promoting automation and intelligence of fault diagnosis in injection molding systems. Similar technical routes can be generalized to other industrial scenarios for quality inspection problems.

Physics-informed DeepONet (PI_DeepONet) is utilized for the reconstruction task of structural displacement based on measured strain. For beam and plate structures, the PI_DeepONet is built by regularizing the strain–displacement relation and boundary conditions, referred to as geometric differential equations (GDEs) in this paper, and the training datasets are constructed by modeling strain functions with mean-zero Gaussian random fields. For the GDEs with more than one Neumann boundary condition, an algorithm is proposed to balance the interplay between different loss terms. The algorithm updates the weight of each loss term adaptively using the back-propagated gradient statistics during the training process. The trained network essentially serves as a solution operator of GDEs, which directly maps the strain function to the displacement function. We demonstrate the application of the proposed method in the displacement reconstruction of Euler–Bernoulli beams and Kirchhoff plates, without any paired strain–displacement observations. The PI_DeepONet exhibits remarkable precision in the displacement reconstruction, with the reconstructed results achieving a close proximity, surpassing 99%, to the finite element calculations.

The discrete element method (DEM) is a vital numerical approach for analyzing the mechanical behavior of elastoplastic wet sand. However, parameter uncertainty persists within the mapping between constitutive relationships and inherent model parameters. We propose a Parameter calibration neural network based on Attention, Retention, and improved Transformer for Sequential data (PartsNet), which effectively captures the nonlinear mechanical behavior of wet sand and obtains the optimal parameter combination for the Edinburgh elasto-plastic adhesion constitutive model. Variational autoencoder-based principal component ordering is employed by PartsNet to reduce the high-dimensional dynamic response and extract critical parameters along with their weights. Gated recurrent units are combined with a novel sparse multi-head attention mechanism to process sequential data. The fusion information is delivered by residual multilayer perceptron, achieving the association between sequential response and model parameters. The errors in response data generated by calibrated parameters are quantified by PartsNet based on adaptive differentiation and Taylor expansion. Remarkable calibration capabilities are exhibited by PartsNet across six evaluation indicators, surpassing seven other deep learning approaches in the ablation test. The calibration accuracy of PartsNet reaches 91.29%, and MSE loss converges to 0.000934. The validation experiments and regression analysis confirmed the generalization capability of PartsNet in the calibration of wet sand. The improved sparse attention mechanism optimizes multi-head attention, resulting in a convergence speed of 21.25%. PartsNet contributes to modeling and simulating the precise mechanical properties of complex elastoplastic systems and offers valuable insights for diverse engineering applications.