ABSTRACT Accurate prediction of the deviations/deformations is highly necessary for the new product development process. It is a tedious multi-tasking activity, in which various aspects need to be considered in the early phase of design. This paper proposes a new constitutive and collaborative framework to model the functional assembly geometry. For complex industrial applications in the multi-plant scenario, several departments work together for a common goal. Often the department’s goal is different and cannot achieve the feat of right for the first time. Hence this paper aimed to integrate the vital departments of the manufacturing industry. Initially, the functional assembly was predicted through Finite Element Analysis (FEA). A multi-layer perceptron type (MLP) artificial neural network was employed to learn the FEA behavior of the assembly. Further, the assembly prototype was practically tested to validate the FEA results, and the obtained data were used to verify the MLP network model. The best trained and tested network model was simulated to predict the near-net geometry considering the functional behavior of the assembly with external loads. The proposed method provides affirmative knowledge while integrating the finite element analysis, testing methods, and neural networks.

中文翻译：

---

通过使用云平台集成设计、检查和测试活动的智能组成和协作框架

摘要 对偏差/变形的准确预测对于新产品开发过程是非常必要的。这是一项繁琐的多任务活动，在设计的早期阶段需要考虑各个方面。本文提出了一种新的本构和协作框架来对功能装配几何进行建模。对于多工厂场景中的复杂工业应用，多个部门为了共同目标而共同努力。往往部门的目标不同，不能第一次实现对的壮举。因此，本文旨在整合制造业的重要部门。最初，功能组件是通过有限元分析 (FEA) 进行预测的。采用多层感知器类型 (MLP) 人工神经网络来学习组件的 FEA 行为。此外，对装配原型进行了实际测试以验证 FEA 结果，并使用获得的数据来验证 MLP 网络模型。考虑到组件在外部负载下的功能行为，模拟了经过最佳训练和测试的网络模型，以预测近网几何形状。所提出的方法在集成有限元分析、测试方法和神经网络的同时提供了肯定的知识。