Innovative data techniques for centrifugal pump optimization with machine learning and AI model

Gaurav Sandeep Dave, Amar Pradeep Pandhare, Atul Prabhakar Kulkarni and Dhananjay Vasant Khankal

PLOS ONE, 2025, vol. 20, issue 6, 1-27

Abstract: In modern centrifugal pump machines (CPM), a data acquisition system encompassing software- hardware interfacing is essential for parameter recording. The quality of recorded data plays a crucial role and directly influences the data transformation phase in machine learning (ML) and deep learning (DL) models. The Dewesoft FFT DAQ system is designed to extract the high-quality data from the CPM based on sensor fusion technology. The data recorded from DAQ system undergoes thorough in-depth analysis, processing & transformation before being incorporated into machine learning (ML) or artificial intelligence models. This paper emphasizes the importance of data cleaning, pre-processing, and applying appropriate methodologies to transform raw data into a valuable resource that can be utilized by ML and AI models. Key techniques include Exploratory Data Analysis (EDA), Data Visualization, and Feature Engineering (FE), which collectively enhance data interpretability. Following these transformations, hypothesis testing validates the data’s integrity, ensuring reliability for subsequent modeling. The validated data is employed to train machine learning classifiers and deep learning algorithms, targeting a 27.25% enhancement in operational efficiency based on F1 score. Additionally, it decreases model training time by 180 seconds, facilitating predictive maintenance of critical performance metrics and minimizing downtime. The assessment of model performance relies on Precision, Recall, and F1 score. This approach leverages recent advancements in data science to derive actionable insights from CPM data, facilitating more informed decision-making and optimization of pump operations.

Date: 2025
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0325952 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 25952&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0325952

DOI: 10.1371/journal.pone.0325952

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().