A new manufacturing control system using Mahalanobis distance for maximising productivity

Primary productivity in the semiconductor manufacturing industry, which includes cycle time, cost and production, depends to a great extent on the capability of manufacturing administrators (MA), particularly with regard to the critical trade-off between cycle time and tool utilization. The Mahalanobis distance (MD) has significance in pattern recognition, and we have found a method to make use of the MD as the core of a manufacturing control system. By using this system, we can easily distinguish deviations from normality in respect of productivity, specify the root cause of the abnormality and decide how to prioritize the problem. As a result, we can efficiently concentrate limited resources on the root cause in the absence of a capable MA, and restore productivity on a minimum timescale.