Self-Validating Knowledge-Guided On-Line Optics and Orbit Correction in an Electron Storage Ring Using a Combined Neural Network and First-Principles Model

75866-The difficult problems being tackled in the accelerator community are those that are nonlinear, substantially unmodeled, and vary over time. These problems would be ideal candidates for model-based optimization and control if representative models could be developed. Such models must be inexpensive to deploy and maintain, and must remain valid throughout the operating region of the system and through variations in system dynamics. This project will develop methodology and algorithms for building high-fidelity mathematical representations of complex nonlinear systems via a combination of first-principles and neural network models. Phase I will develop combined first-principles and neural networks models for optics and orbit correction on two challenging problems in the electron and positron storage rings at a national laboratory; the models will have a direct impact on luminosity and beam lifetime, factors that influence research efficiency in such a facility. Algorithms for self-validation of the combined models will be developed, and the diagnostic value of these algorithms will be determined. Commercial Applications and Other Benefits as described by the awardee: In addition to the application for high energy accelerators, the modeling of complex nonlinear systems with potentially varying dynamics should be applicable to the process industries.