Development of Predictive Maintenance Technology for Wafer Transfer Robot using Clustering Algorithm

Abstract

Semiconductor equipment is closely related to production yield and should be periodically inspected to maintain the performance. However, traditional equipment maintenance methods have difficulty responding to sudden failures and economic losses due to unpredicted situation. Recently, predictive maintenance technologies have been actively studied, with many advantages, such as the ability to predict the right time of equipment maintenance in advance to ensure system stability and maximize return on capital. Economic gains and increased equipment stability are particularly important in the semiconductor industry. In this paper, we developed a predictive maintenance technique for wafer transport robots. A technology was developed to predict the failure of the robot by tracing the median values of clusters of the frequency data, which were acquired using acceleration sensors attached on the handler of the robot. By applying this system to the wafer transport robot, the simulation results showed that the time of repair can be predicted in real time.