In a dynamic manufacturing environment, it may not be adequate to only look for known process problems, but also important to uncover and react to new, previously unseen patterns as they emerge.
Univariate and linear multivariate statistical process control methods have traditionally been used in the semiconductor industry to detect anomalies. With increasing equipment, process, and product complexity, multivariate anomalies that also involve significant interactions and nonlinearities may be missed by these more traditional methods.
We will demonstrate a method for identifying complex anomalies using a deep learning autoencoder. Once the anomalies are detected, their fingerprints are generated so they can be classified and clustered, enabling investigation of the causes of the clusters. As new data streams in, it can be scored in real time to identify new anomalies, assign them to clusters, and respond to mitigate potential problems.
These tools are no longer the exclusive province of data scientists. With today’s analytics platforms, they can be utilized by virtually all engineers.