Machine-Learning Fast Automatic Shifter Tuning

To address the Air Force need for an improved thermal characterization process for ferrite phase shifters used in phased array antennas, Physical Optics Corporation (POC) has developed a new Machine-Learning Fast Automatic Shifter Tuning (M-FAST) system. The current method to characterize ferrite phase shifters used in phased array antennas during the overhaul process at OC-ALC requires a minimum of 16 hr. In Phase I, POC demonstrated the feasibility of M-FAST by measuring and characterizing a phase shifter. POC-developed intelligent log data analysis tools in M-FAST reduce the number of response curves required to fully characterize each phase shifter from several dozen to less than ten. As a result, M-FAST can measure more phase shifters in a shorter time and with fewer measurement points, meeting the Air Force requirements for this topic. In Phase II, POC will optimize the M-FAST algorithm to reduce the cost and improve the quality of repairs that require the characterization process and will perform validation testing and performance evaluation of the M-FAST system. During Phase III, POC intends to commercialize M-FAST for use by commercial and/or Government, overhaul entities, and DoD user/depot facilities.