OptNeTol: Integrated, optimization-based parameter and tolerance design

使用优化方法进行最优公差分配，对开发低成本、高质量的产品具有重要意义。通过构造合适的优化问题并借助强大的算法进行求解，可以在确保满足对健壮性、功能和美观性的严格要求的同时，以充分挖掘成本潜力的方式分配公称尺寸和公差值。通过这种方式，基于优化的方法相继取代了纯粹的定性方法，如简单的经验法则或众所周知的定理，如定义公差尽可能宽，但尽可能紧，从而创造了无可争议的竞争优势。在前面的资助阶段，为公差优化在工业环境中的生产性应用奠定了重要的基础。然而，它还不能有效地帮助回答与实践相关的问题。这主要是由于高效、有效的优化方法的不足，跨学科方法之间缺乏联系，以及缺乏对实际问题的关注，因此，本研究项目的目的是使产品开发人员能够将成本最优的零件尺寸和公差分配到实际应用案例中。这意味着跨学科和高度专业化的公称尺寸和公差优化方法在很大程度上消除了实际应用的障碍。在此背景下，开发了公称尺寸和公差优化的创新方法，解决了到目前为止还没有考虑到的方面，但对于实际应用来说是绝对必要的，例如几何公差或对缺失或不确定成本信息的处理。然而，由于优化问题只能通过强大的、基于抽样的优化算法来解决，因此本研究项目的重点也在于开发和组合各种方法，以显著提高确定有效优化结果的效率，例如自适应抽样和代理建模，包括统计容差分析方法。最后，单独开发的方法将在基于共同的跨学科知识库的软件原型中联系起来，其适用性将在用户研究中进行评估。这确保了对优化没有深入了解的产品开发人员在尽可能自动地定义和解决实际问题时得到支持。因此，目前阻碍从单纯公差优化的专家工具向集成的、辅助优化的公差管理转变的公开研究空白最终被填补。

The usage of optimization methods for optimal tolerance allocation significantly contributes to the development of least-cost and high-quality products. By formulating suitable optimization problems and solving them with the aid of powerful algorithms, nominal dimensions and tolerance values can be allocated in a way that the cost potential is fully exploited while ensuring that the stringent requirements on robustness, function and aesthetics are fulfilled. In this way, optimization-based methods successively replace purely qualitative approaches, such as simple rules of thumb or well-known theorems like "define tolerances as wide as possible, but as tight as necessary", and thereby create an undisputed competitive advantage.In the preceding funding phase, an important basis for the productive use of tolerance optimization in the industrial environment was established. However, it cannot effectively contribute to answering practice-relevant questions yet. This is mainly due to the deficits of efficient, valid optimization methods, the missing link between the interdisciplinary methods and their lack of focus on practical problems.Therefore, the aim of this research project is to enable the product developer to assign cost-optimal part dimensions and tolerances to practically relevant application cases. This implies that the obstacles to the practical application of the interdisciplinary and at the same time highly specialized methods of nominal dimension and tolerance optimization are largely eliminated. In this context, innovative methods of nominal dimension and tolerance optimization are developed addressing aspects that have not been considered so far, but are absolutely necessary for practical use, such as geometrical tolerances or the handling of missing or uncertain cost information. However, since the optimization problems can only be solved by powerful, sampling-based optimization algorithms, the focus of this research project is also on the development and combination of methods to significantly increase the efficiency in the determination of valid optimization results, such as adaptive sampling and surrogate modelling, including statistical tolerance analysis methods. Finally, the individually developed methods will be linked in a software prototype based on a common, interdisciplinary knowledge base and their applicability will be evaluated in user studies. This ensures that product developers without in-depth knowledge of optimization are supported in defining and solving practical problems as automatically as possible. Thus, currently open research gaps preventing the change from an expert tool for mere tolerance optimization to an integrated, optimization-aided tolerance management are finally closed.