Design and Deployment of a Data Lake at a Pilot Plant Scale for a Smart Electropolishing Process
Bibliographic record
Abstract
In order to remain competitive and satisfy the demands of today’s customers in a timely manner, manufacturing industries are embracing the Industry 4.0 philosophy where automation is pushed beyond robotics to new technologies emerging from data science and artificial intelligence. The aim is to reduce time spent on none added value tasks and help learning from past experience in order to enhance efficiency and quality of manufacturing processes. Traditional industries, such as electropolishing, need to find ways to automate their, often heavily artisanal-based techniques and develop an intelligent network of machines and processes taking advantage of information and communication technology such as Big Data, IoT (Internet of Things), or Artificial Intelligence (AI). This digital transition can be realized through the application of an IIoT (Industrial Internet of Things) platform that constructs a massive, sophisticated information network of interconnected sensors, equipment, and processes known as cyber-physical systems. Within this network, large amounts of data (for example process bath attributes such as temperature or viscosity and part characteristics such as roughness or brightness) can be collected automatically via sensors and through user-friendly applications from manual measurements and observations. All data are uploaded automatically into a cloud-based data storage system. In order for this collected information to be useful, the data needs to be processed to allow pattern discovery and extraction of useful information regarding the system performance, probable faults in the process, and product quality. Besides others, machine learning algorithms play a key role in extracting useful information. Classification and processing of such massive, diverse, and rapidly arriving data sets are known to be challenging. As a result, the concept of data lake has arisen in the last decade as an appealing and cost-effective approach for companies to manage large amounts of data. It consists of a large repository of datasets designed to transform raw and unstructured data into structured, usable information to allow further processing. A data lake, organized typically in four layers (ingestion, distillation, processing, and insights layers), stores both old and near real-time data in one location for initial assessment, with comprehensive data organization, analysis, and visualization being performed only when necessary 1,2. This promotes agility by allowing data to be accessed by everyone in the company. 2 In this work, a data lake is designed and implemented in conjunction with a pilot plant to demonstrate how in the electropolishing process of stainless-steel samples in an aging electrolyte, data can be collected and organized for further processing using machine learning techniques in order to optimize the process and part quality based on the data analysis results. References: N. Miloslavskaya and A. Tolstoy, Procedia Comput. Sci., 88, 300–305 (2016). H. Fang, in 2015 IEEE International Conference on Cyber Technology in Automation, Control, and Intelligent Systems (CYBER),, p. 820–824 (2015). Figure 1
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.015 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".