抄録
In times of smart products and systems, the Internet of Things (IoT) plays an increasingly important role. IoT combines the digital world (internet) with the physical world (sensors, actuators, robots, smartphones, connected cars etc.). The autonomous operation and remote control of smart systems (e.g. smart home, production hall or line) requires efficient and specially reliable actuators and control mechanisms. Shape memory actuators are particularly suitable for this application due to their properties as they are lightweight, small, energy-efficient and can also be used as sensors at the same time. Many shape memory actuators have been developed for various applications over the past years. Despite great interest, there are no standardized test programs available. The complexity of the shape memory technology is a major challenge in testing fatigue and degradation behavior of components to determine reliability. This article presents fatigue test results of a laboratory test rig of a case study for a shape memory actuator, including all boundary conditions and test requirements. The measurement data consists of different parameters e.g. the stroke of the actuator, the electrical voltage and current (to activate the actuator) as well as the ambient temperature. Since the study comprises only a few prototypes, parametric methods are not suitable for a comprehensive evaluation, therefore parameter-free methods are used as well. The analysis regarding the description of dependencies between the recorded signals and the detection of degradation of the shape memory actuators is discussed in detail. The main objective is the development of a prognosis algorithm in order to be able to predict the failure behaviour of the actuators at an early stage. The methodical approach includes various methods and procedures, which are applied in a logical order. The statistical analytics used in this study are focusing on nonparametric significance tests, such as the Levene’s test and the u-test by Wilcoxon and Mann-Whitney. Further methods are the correlation analysis and the regression analysis as well as multivariate 3D-plots. The fundamentals of shape memory alloys, as well as the used statistical nonparametric methods, are described briefly. Finally, the realization (application of the analysis methods) based on the real test rig data of a case study consisting of 18 actuators is shown and discussed in detail.
