Statistical analysis of machinery variance by python

Abstract

Based particularly on data technologies, information is rapidly evolving in engineering. In mechanical engineering, maintenance is benefiting the most from data innovations, the reduction of maintenance costs, and the improvement of system availability. Modern reliability engineering targets uncertainty with advanced statistics and data science. This paper explores maintenance descriptive analytical techniques to determine whether variations between porosity percentage in batches from two similar machines are due to randomness, suggesting technical issues on machinery. The finding is backed by statistical analysis, hypothesis testing, and data mining. Python 3 was used following best practices of data science and data visualization as the main toolset to explore the working data, execute the proposed statistical analysis, and make conclusions. The findings were promising reinforcing the importance of advanced statistics for reliability studies. As a conclusion, we failed to reject the null hypothesis suggesting that there was no apparent difference between sample means. With a confidence of 95%, the difference is explained by natural variations and randomness inherent to the fabrication process.