An efficient Top-Down Redesign Approach: Translating System Specifications into Component Requirements

Complex mechatronic systems generally consist of an interconnection of multiple subsystems or components, possibly involving multi-physics aspects. To achieve high-performance and high-precision systems, typically, first, system engineers define design specifications on the interconnected system level. Then, these specifications should be translated into requirements for each component, which generally is a far from trivial task. Based on these requirements, each individual component can then be (re)designed and subsequently developed. After the development of each component, they are individually validated on their respective component requirements. If a component is validated and it does not meet the requirements, its design is generally updated until it meets those requirements, which is relatively cheap, since this process can be done entirely on component-level. Finally, the complete system is integrated and the overall system specifications are validated. If the (interconnected) system does not satisfy these specifications, the component requirements, and, consequently, the component designs themselves, need to be adapted, which generally is a highly costly procedure. Therefore, for an efficient (re)design process, it is crucial that, after translating system specifications into component requirements, satisfaction of these component requirements indeed guarantees satisfaction of the interconnected system specifications.

In this work, we assume that a model of the mechatronic system and its components already exists. We introduce a top-down redesign approach which, given specifications on the overall, interconnected system, can determine requirements for each individual component. Here, both system and component requirements are given in terms of maximally allowed change in the dynamics of the existing system model. The requirements are distributed to the components such that components that have less effect on the specifications of the system as a whole, can be redesigned/modified more. Furthermore, this approach allows that modifications to components can be made to improve the performance, precision, monitoring and reliability of a system, or reduce the cost, in a completely modular approach, i.e., only satisfying and optimizing within local component requirements. Crucially, this approach provides the guarantee that if the individual components meet their respective requirements, the system specifications are also satisfied.

The method is inspired by earlier work of the authors [1]. In [1], we reformulate a model reduction problem into a robust performance analysis framework, allowing for tools such as μ-analysis from the field of robust control to be applied. By modelling the system specifications on the system dynamics as a robust performance criterion, it allows us to use this approach to provide maximally allowed design modifications on component level that satisfy the criterion (i.e., the system specifications) by modelling these component modifications as uncertainties.

We show the effectiveness of this redesign approach on an illustrative case study, inspired by an industrial wire bonder model (see Figure 1). In this case study, we investigate the maximum allowed mass of an additional sensor to different locations on individual components, given a bound on the influence this sensor can have on the overall system dynamics at the most relevant frequencies.