Mudassar Hussain
CNC End e??ector tools wear SOH prediction: Big Data.
Rel. Daniele Apiletti. Politecnico di Torino, Corso di laurea magistrale in Data Science And Engineering, 2023
|
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (2MB) | Preview |
Abstract: |
CNC End effector tools wear SOH prediction: a big-data approach Pertaining to the high cost of maintenance of various machines in the industrial sector, there is a growing need of efficient and cost saving techniques to predict the state of health(or remaining useful life of a machine).It is obvious that applying efficient techniques can lead to a competitive advantage in the industrial sector. Our thesis extends the work done previously on the MorePRO project based on the estimation of the state of health of CNC machines. The contrast with the previous work is to focus on cloud based computing rather then previously performed analysis with edge computing which is related to the data collected in the local environment from the machine. Both of these techniques have there own pros and cons. Mainly edge computing being local and thus has low latency while cloud computing encompasses a general analysis consequently leads to better accuracy. The goal is to understand the model pertaining the CNC machine, identifying the useful features that need to be extracted during the ETL process and then apply e??cient machine learning algorithm on the extracted dataset using a cloud based approach, deploying the model using real time streaming. Main steps include loading dataset and performing analytic in a spark based environment. The data exploratory analysis is performed using python based library along with visualization. Various regression models are trained and validated using test data. Finally designing pipeline integrating spark streaming with Apache Kafka Our work focuses on performing regression task to predict the state of health (assumed as a continues variable). Since the data is coming from various models, we need to concatenate these data sets and then perform analysis on a cloud based platform given that the volume of data collected from various models requires more memory and consequently high computation power. We would like to mention that these computations are performed on the cluster provided by Politecnico di Torino. |
---|---|
Relators: | Daniele Apiletti |
Academic year: | 2022/23 |
Publication type: | Electronic |
Number of Pages: | 78 |
Subjects: | |
Corso di laurea: | Corso di laurea magistrale in Data Science And Engineering |
Classe di laurea: | New organization > Master science > LM-32 - COMPUTER SYSTEMS ENGINEERING |
Aziende collaboratrici: | Brain technologies |
URI: | http://webthesis.biblio.polito.it/id/eprint/26877 |
Modify record (reserved for operators) |