A Cloud-to-Edge Approach to Support Predictive Analytics in Robotics Industry

Abstract

Data management and processing to enable predictive analytics in cyber physical systems holds the promise of creating insight over underlying processes, discovering anomalous behaviours and predicting imminent failures threatening a normal and smooth production process. In this context, proactive strategies can be adopted, as enabled by predictive analytics. Predictive analytics in turn can make a shift in traditional maintenance approaches to more effective optimising their cost and transforming maintenance from a necessary evil to a strategic business factor. Empowered by the aforementioned points, this paper discusses a novel methodology for remaining useful life (RUL) estimation enabling predictive maintenance of industrial equipment using partial knowledge over its degradation function and the parameters that are affecting it. Moreover, the design and prototype implementation of a plug-n-play end-to-end cloud architecture, supporting predictive maintenance of industrial equipment is presented integrating the aforementioned concept as a service. This is achieved by integrating edge gateways, data stores at both the edge and the cloud, and various applications, such as predictive analytics, visualization and scheduling, integrated as services in the cloud system. The proposed approach has been implemented into a prototype and tested in an industrial use case related to the maintenance of a robotic arm. Obtained results show the effectiveness and the efficiency of the proposed methodology in supporting predictive analytics in the era of Industry 4.0.

Data mining in predictive maintenance systems: A taxonomy and systematic review

“…In this way, the aim is to simulate behaviors that could occur in a real system to which there is not access or which has not yet failures that can be used to train the models. In this line, there are many works that collect vibration signals from different experimental rotors (Cakir et al, 2021; Li et al, 2019; Liang et al, 2020; Liao et al, 2016; Pang et al, 2020; Qian et al, 2019; Satishkumar & Sugumaran, 2017; Song et al, 2021; Una et al, 2017; Wang et al, 2018; Wang, Zhang, et al, 2020; Yang, Lei, et al, 2019; Zhang, Li, Wang, et al, 2019), fans (Sampaio et al, 2019; Xu et al, 2021; Zenisek, Holzinger, & Affenzeller, 2019), centrifugal pumps (Hu et al, 2020), air compressors (Cupek et al, 2018) or industrial robots (Panicucci et al, 2020). Some works (Venkataswamy et al, 2020; Zenisek, Holzinger, & Affenzeller, 2019; Zenisek, Kronberger, et al, 2019) generate synthetic data from a mathematical approach that models the behavior of their problem, using for that specialized software in simulation.…”

“…The most common method to find correlations is the Pearson correlation coefficient, which focuses only on linear relationships. This is a common method for feature selection in PdM, selecting only those variables that are most strongly related to the target variable (Hsu et al, 2020; Pałasz & Przysowa, 2019; Quatrini et al, 2020), or reducing the number of input variables used for prediction by avoiding those that are redundant because they are strongly correlated with others (Panicucci et al, 2020; Shamayleh et al, 2020). Another feature selection method called Mathew Correlation Coefficient is used in Kolokas et al (2020).…”

“…In Ullah et al (2017), Uhlmann et al (2018), and Ruiz‐Sarmiento et al (2020), the statistics are combined with the original features in order to provide additional information. In many PdM works, due to the temporal relationship of the data, it is common to perform statistics operations by working cycles (Baptista et al, 2021; Cerquitelli et al, 2021; Kolokas et al, 2020; Orrù et al, 2020; Panicucci et al, 2020; Quatrini et al, 2020; Susto et al, 2015), in such a way that new features are computed at each time interval. Another approach is found in Zenisek, Holzinger, and Affenzeller (2019), where time series are preprocessed by generating time lagged values which are referred to extract new features that historically saw the last n values in a series to a new input vector.…”

Extreme Communication in 6G: Vision and Challenges for ‘in-X’ Subnetworks

“…Also, the AP in each subnetwork can collect statistics and KPIs of the supported control loops. For example, it can transfer the statistics of the jitter patterns of the received measurement from the sensors, which can be processed by the local or an edge cloud server [35]. Such server can make use of machine learning techniques for identifying potential anomalies in the behavior of the robots, and eventually take actions, e.g., stopping robot activity if it is foreseen it can harm production efficiency or create hazard.…”

Energy-Based Prognosis of the Remaining Useful Life of the Coating Segments in Hot Rolling Mill

“…Predictive maintenance assumes that the monitored machine parts go through a measurable process of degradation, hence enabling the estimation of temporal windows for carrying out repair operations [2,13]. PdM comes to benefit from technological advances and predict the RUL of components through degradation measuring.…”