The term robustness (lat. Robustus , from robur hardwood, oak) describes the ability of a system to withstand changes without adapting its initially stable structure. In most cases, it makes sense to specify what the system is robust against (e.g. against changes in the ambient temperature or against incorrect operation).
Robustness in various disciplines
In computer science and software development , the term “robustness” means the property of a process to function reliably even under unfavorable conditions. Robustness, also known as "fault tolerance", is one of the quality criteria for software.
Examples of appropriate precautions are the prevention of undefined states and "system crashes" (e.g. through complete and detailed evaluation of response codes after the execution of subroutines or system calls) and in particular the interception of incorrect user or data entries (such as invalid commands / function codes , incorrect ones Formats in data fields, etc.).
As far as possible, a “spectrum of sensible reaction options, depending on the situation” should be defined and applied, which, depending on the expectations, can mean a high implementation effort.
Nevertheless, one hundred percent robustness will not be achievable, for example if required components of the system software are missing or do not work correctly. But even in such cases, a computer program can u. U. generate an error message that is as informative as possible and terminate itself in a controlled manner.
In industry, the term "robust production process" is used. For the automotive industry there is a VDA volume on this in the series “The common quality management in the supply chain” with the title “Product manufacture and delivery, robust production process”. According to this, a robust production process is characterized by the fact that it is insensitive to undesirable influencing variables and ensures that production is on schedule and on demand with excellent quality while maintaining the planned economic outlay. The definition of the “robust production process” is also the definition of the “guard rails” that describe the path to implementation.
In analysis or diagnostics , the robustness of an analytical system allows a certain variability of the sample to be analyzed (e.g. sample pretreatment not required) and / or other defined physical parameters during the measurement process and still delivers reproducible and standardized results.
In inferential statistics , robustness means that e.g. For example, a test works reliably even if the conditions are violated (e.g. no normal distribution , too small a sample ) and the errors of the 1st and 2nd type only change slightly. In the case of a lack of robustness, violations of the prerequisites lead to errors of type 1 or type 2 and lead either to progressive (incorrect rejection of the null hypothesis ) or conservative decisions (incorrect retention of the null hypothesis). For more information see Robust Estimation Methods , M-Estimators , Breakpoints , Outliers .
Robustness of a biological system is the evolutionary persistence of a certain feature or property in a system under disturbance or uncertainty conditions. Robustness in embryonic development is channeling .
The opposite of robustness is tenderness, fragility or fragility as an unstable, but easily vulnerable constitution / constitution.
- Wieland, A., Wallenburg, CM, 2012. Dealing with supply chain risks: Linking risk management practices and strategies to performance. International Journal of Physical Distribution & Logistics Management, 42 (10). In the English original: "the ability of a [system] to resist change without adapting its initial stable configuration"
- TU Harburg Lecture Software Engineering  (PPT file; 2.6 MB)
- Cf. Christof Nachtigall and Markus Wirtz: Probability Calculation and Inference Statistics. Juventa Verlag, Weinheim 2006, 4th edition, ISBN 978-3-7799-1052-7 , p. 215 f.
- Hiroaki Kitano: Biological robustness . In: Nature Reviews Genetics . 5, No. 11, 2004, pp. 826-37. doi : 10.1038 / nrg1471 . PMID 15520792 .
- CH Waddington: canalization of Development and the Heritance of Acquired character. In: Nature, 1942, 3811, pp. 563-565.