MACHINE LEARNING BASED LOGANALYSIS FOR AUTOMATED ANOMALY DETECTION

Many sensors are used in a single production process, making it difficult to pinpoint the exact source of a problem. More than one process cycle is required to make a semiconductor wafer. There are many cycles in this process, and it is difficult to spot abnormalities in time; thus, the process continues until it is complete. The cost of producing these wafers is high, and a process failure can have a significant impact on both time and money. As a result, anomaly detection in semiconductor production can benefit greatly from machine learning. A manufacturing facility may interrupt the operation and fix the problematic equipment if irregularities in the production process could be discovered or predicted sooner. As a result, semiconductor producers would see an improvement in process yield and a reduction in expenses.