Abstract
Abstract
At present, composites are formed by multi-phase materials with reinforcement and matrix, which can produce a discontinuous interface layer between two phases. Therefore, surface modification methods are proposed to enhance the performance and the composite properties. However, the evaluation of these properties primarily relies on manual detection and this may cause mistakes in some situations. In this work, an automatic evaluation model is proposed to achieve the interfacial properties estimation. The neural network consists of multiple layers of neural cells and the evaluation is simulated by the prediction process with enough training iterations. Initially, the neural network model is trained with tremendous composite properties with different surface modification methods and the evaluation results are labelled. Further, the inner parameters are set and the convergence condition is controlled to obtain an acceptable evaluation model. Finally, our model is compared with existing evaluation methods. From our extensive experiments and analysis, it can be concluded that our model can realize the evaluation for the performance of surface modification in interfacial properties.