Optimal Model-Based Fault Estimation and Correction for Particle Accelerators and Industrial Plants Using Combined Support Vector Machines and First Principles Models

One common theme in the operation of storage rings and linear accelerators is the need to precisely control the particle beams over long periods of time, with minimum beam loss and stable beam parameters. Timely estimation of deviations from optimal performance in these complex systems, and the ability to identify corrective measures in response to estimated parameter deviations, has been the subject of extensive research over the past four decades. This project will develop a new framework for building parsimonious models that are suited for diagnosis and fault estimation of complex technical systems. The framework will perform an optimal, nonlinear principal-component analysis of the large-scale input space and create a lower-dimension feature space in which fault-estimation results can be effectively presented to operational personnel. The framework will use Support Vector Machines (SVMs) to model potentially time-varying parameters of a First-Principles (FP) description of the process. Unlike prior approaches to modeling, the combined SVM/FP model will be built (i.e., model parameters trained) with extremely efficient convex optimization techniques. The trained models then can be used for online/offline diagnosis and fault estimation in realistic applications with large number of inputs and outputs parameters. Commercial Applications and other Benefits as described by the awardee: Beyond applications in particle accelerators, the high fidelity and cost benefits of a combined model-based fault estimation/correction system should attract customers from a wide variety of commercial and scientific industries, including the process industry (polymer manufacturing, power generation, bio-fuel production, food processing, cement, and pulp and paper), the pharmaceutical industry, and any other industry in which high-accuracy and high speed fault estimation systems are desirable (e.g. avionics, underwater submarines, and embedded estimation of faults in Unmanned Aerial Vehicles). The implications in terms of lost revenue from costly industrial processes, operation of large-scale public works projects and the volume of the published literature on this topic clearly indicates the significance of the problem. Applications range from manufacturing industries (integrated circuits, automotive, etc.), to largescale chemical plants, pharmaceutical production, power distribution grids, and avionics.