ML Clustering Analysis of LAPD Machine State Information for Identifying Correlations

The LAPD’s machine state information (MSI) is composed of various diagnostics and sensors which provides information about machine configuration and plasma state. This MSI can then be correlated to discern trends in the behavior of the LAPD plasmas. Manually performing this correlation process is very complex because of the many unique plasma configurations recorded. We utilize python-based clustering algorithms to identify correlations and trends between discharge current, discharge voltage, gas pressure, and magnetic field profile across a variety of LAPD configurations. Initial results from two dimensional clustering imply some correlation between MSI signals but the trends remain unclear. Either higher dimensional clustering is required to identify the complex correlations between various MSI signals or more granular meta-clustering is required to identify data trends within individual clusters. Initial results as well as results of higher dimensional clustering and meta-clustering will be presented.