Predictive Maintenance Using Sensor Data Processing

The chapter explores the various types of equipment failure, techniques to discover the root cause, and different approaches to find the solution to predict the event of the failure. Moreover, this work addresses the equipment failure prediction on the assembly line of truck wheel manufacturing plant. The step-by-step process of the wheel production includes one after another usage of the machines. Any workstation failure leads to the idle time of the subsequent machines and reduction in the working efficiency of the labor. The sensor-based data of the machine is collected using the process automation. Historical data is used to predict the remaining useful life. K-Nearest Neighborhood technique has given the accuracy of 96.40%. The Decision Tree with Gini Index and Entropy have has achieved the 96% accuracy in the machine state prediction. The multi-regression based model developed to calculate the values of coefficient and intercept as learning parameters to be useful in the further deployment.