Due to development of novel and more efficient energy storage systems we bear witness to the dawn of a new era of mobile systems. They have become sophisticated in terms of hardware components and software applications which have made it possible to develop integrated solutions for a large number of imaginable applications ranging from electric vehicles all the way to fully autonomous systems operating in a wide variety of ecosystems, e.g., service, surveillance or bio-inspired robots. Generally it is expected that a mobile system exhibits a sufficient degree of autonomy in the sense of energy availability such that it at least accomplishes the mission objectives for which it is intended. Nevertheless, such autonomy, is influenced to a large extent by the remaining energy that can be retrieved from its energy storage system and by the environment conditions in which the system operates. Assessing the reliability of a mission requires using systems internal and external situational awareness to determine if the available energy at least meets the energy needs demanded by the future operation of the mobile system in order to determine its remaining useful life (RUL). Having this information as soon as possible may allow the decision maker to apply a contingency plan to intervene and reconfigure the mission execution strategy in order to improve the probability of success, in those situations in which the system becomes incapable of achieving the original mission objectives. Numerous studies have been published for assessing mission reliability and estimating the RUL of mobile systems. However, they deal with structured environment conditions and thus with relatively deterministic loads. Moreover, these approaches neglect the inherent uncertainty which stems from multiple sources such as the lack of knowledge about the true energy available in the mobile system, the noise introduced by sensors or the randomness of the operation environment, just to mention a few. The approach presented in this work is built around the belief that the RUL estimation is formulated as an uncertainty propagation problem. Accordingly, to estimate the RUL multiple sources of uncertainty involved in its estimation are first characterized and then propagated with the aim of computing their combined effect, expressed in terms of a probability density function. The approach developed here achieves this estimation in a Monte-Carlo fashion in which several RUL realizations are simulated in order to accurately estimate its entire probability distribution. The aim of this work is therefore devoted to develop a solution capable of estimating the RUL with application to energy-constrained mobile systems operating in stochastic environments.
The different versions of the original document can be found in:
Are you one of the authors of this document?