Optimizing Manufacturing Process Through Advanced Analytics and Machine Learning Techniques

Machine failure is prevented before it happens in the industrial sector through a process called Maintaining Predictiveness. Manufacturing companies need to periodically inspect their machines and use sensors to assess their level of intelligence during the time these actions support constant output, which subsequently change supports the maintenance of product quality. Additionally, they reduce costs by keeping the equipment in optimal condition and by preventing costly replacements. There are various approaches available to the manufacturing sector for implementing predictive maintenance, whereas each has limitations and drawbacks. For example, these methods struggle to handle large amounts of machine data and deliver the highest degree of accuracy. They also cannot be used in dynamic environments. At present, machine learning is a technique that is showing its efficiency in almost every sector. They are especially helpful in predictive maintenance in the manufacturing industry, whereas they perform every purpose effectively. The A141 2020 dataset, which contains information about the machinery and the time it has taken, has been utilized to implement two distinct algorithms for machine learning in this research for predictive maintenance: the random forest algorithm and the Divide and Conquer Kernel Support Vector machine learning algorithms. Prior to the data being implemented with the algorithms, the preprocessing step addressed the issue of missing attribute values in the dataset using the K-Nearest Neighbor Algorithm. Then, the data was implemented with these two algorithms, and the performance metrics, including accuracy, precision, recall, and F1 score, were used to compare the results. Applying this set of data for predictive maintenance, the outcome clearly indicates that the DCKSVM algorithm performs better than the Random Forest Algorithm.