The penetrating radiography provided by the Dual Axis Radiographic Hydrodynamic Test (DARHT) facility is a key capability in executing a core mission of the Los Alamos National Laboratory (LANL). Historical data from the two DARHT Linear Induction Accelerators (LIAs), built as hdf5 data structures for over a decade of operations, are being used to train machine learning models to assist in beam tuning. Adaptive machine learning (AML) techniques that incorporate physics-based models are being designed to use noninvasive diagnostic measurements to address the challenge of predicting the radiographic spot size, which depends on the time variation in accelerator performance and the density evolution of the conversion target. Pinhole collimator images recorded by a gamma ray camera (GRC) provide a direct measurement of the radiograph imaging quality but are not always available. A framework is being developed to feed results of these invasive measurements back into the accelerator models to provide virtual diagnostic measurements when these measurements are not available.

The Dual-Axis Radiographic Hydrodynamic Test Facility (DARHT) uses a spinning wheel debris blocker as a crucial machine protection system to prevent target debris from the electron to X-ray conversion process from traveling upstream and damaging the accelerator. The spinning wheel in use on DARHT Axis-I consists of two spinning disks normal to the beamline, each with an open slit that crosses the beamline at frequencies of 50 Hz and 40 Hz, creating an opening at a beat frequency of 8 Hz allowing the electron beam to pass through and shut behind it. In this poster, we present steps taken to improve the reliability and performance of the spinning wheel, which include replacing legacy and custom diagnostic components with off-the-shelf hardware. We also present the challenges and solutions in testing and deploying these upgrades without disrupting operation of the accelerator.