Real-Time Manufacturing Drilling Operations Analysis by Utilization of Data-Fusion

Abstract

In mining and construction operations, the protection, safety and machinery's lifetime hold a crucial concern that can impose unwelcoming costs on the projects. The motivation behind this work is to deliver a model capable of addressing these apprehensions besides managing the potential risks and costs of the types of machinery. The presented model in this article aims to increase the quality and reliability of the products and their operations by utilizing sensor information for real-time prediction and categorization of drilling operations. This model works based on the time analyses on the sensory fused data. We applied the model on the three-axis acceleration and angular velocity signals (generated from a simulated system) to extract features and categorize three different rock drilling operations. For each operation, we measured the Median Absolute Deviation (MAD) and dynamic range parameters of the acceleration signals. In addition, we succeeded to calculate the Root Mean Square (RMS) parameter as a feature from angular velocity signals. The obtained results in this study approve the real-time prediction and categorization potential of the introduced approach for the different rock drilling operations. However, the limitation of this work can be the source of the data which is originating from the simulated normal operations. As an extending future work in future publications, we will include the faulty operation data, the real data from measurements and present data analysis of abnormal operations.