Machine learning (ML) (a subset of artificial intelligence that focuses on autonomous computer knowledge gain) is actively being used across many domains, such as entertainment, commerce, and increasingly in industrial settings. The wide applicability and low barriers for development of these algorithms are allowing for innovations, once thought unattainable, to be realized in an ever more digital world. As these innovations continue across industries, the manufacturing industry has also begun to gain benefits. With the current push for Smart Manufacturing and Industrie 4.0, ML for manufacturing is experiencing unprecedented levels of interest; but how much is industry actually using these highly-publicized techniques? This paper sorts through a decade of manufacturing publications to quantify the amount of effort being put towards advancing ML in manufacturing. This work identifies both prominent areas of ML use, and popular algorithms. This also allows us to highlight any gaps, or areas where ML could play a vital role. To maximize the search space utilization of this investigation, ML based Natural Language Processing (NLP) techniques were employed to rapidly sort through a vast corpus of engineering documents to identify key areas of research and application, as well as uncover documents most pertinent to this survey. The salient outcome of this research is the presentation of current focus areas and gaps in ML applications to the manufacturing industry, with particular emphasis on cross domain knowledge utilization. A full detailing of methods and findings is presented.
To facilitate root cause analysis in the manufacturing industry, maintenance technicians often fill out "maintenance tickets" to track issues and corresponding corrective actions. A database of these maintenance-logs can provide problem descriptions, causes, and treatments for the facility at large. However, when similar issues occur, different technicians rarely describe the same problem in an identical manner. This leads to description inconsistencies within the database, which makes it difficult to categorize issues or learn from similar cause-effect relationships. If such relationships could be identified, there is the potential to discover more insight into system performance. One way to address this opportunity is via the application of natural language processing (NLP) techniques to tag similar ticket descriptions, allowing for more formalized statistical learning of patterns in the maintenance data as a special type of short-text data. This paper showcases a proof-of-concept pipeline for merging multiple machine learning (ML) and NLP techniques to cluster and tag maintenance data, as part of a broader research thrust to extract insight from largely unstructured natural-language maintenance logs. The accuracy of the proposed method is tested on real data from a small manufacturer.
Manufacturing systems are becoming increasingly complex as more advanced and emerging technologies are integrated into the factory floor to yield new processes or increase the efficiency of existing processes. As greater complexity is formed across the factory, new relationships are often generated that can lead to advanced capabilities, yet produce unforeseen faults and failures. Industrial robot arm work cells within the manufacturing environment present increasing complexity, emergent technologies, new relationships, and unpredicted faults/failures. To maintain required levels of productivity, process quality, and asset availability, manufacturers must reconcile this complexity to understand how the health degradation of constituent physical elements and functional tasks impact one another through the monitoring of critical informative measures and metrics. This article presents the initial efforts in developing a novel hierarchical decomposition methodology. The innovation in this method is that it provides the manufacturer with sufficient discretion to physically deconstruct their system and functionally decompose their process to user-defined levels based upon desired monitoring, maintenance, and control levels. This enables the manufacturer to specify relationships within and across the physical, functional, and information domains to identify impactful health degradations without having to know all possible failure modes. The hierarchical decomposition methodology will advance the state of the art in terms of improving machine health by highlighting how health degradations propagate through the relationship network prior to a piece of equipment compromising the productivity or quality of a process. The first two steps of the methodology, physical decomposition and functional decomposition, are defined in detail and applied to a multi-robot work cell use case.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.