Accelerator diagnosis and control by neutral nets

It is suggested that neural nets (NN) provide a good metaphor for large complex systems (LCSs). It can be argued that NNs are logically equivalent to multiloop feedback-forward control of faulty systems and therefore provide an ideal adaptive control system. Thus, while AI (artificial intelligence) may be appropriate for maintaining a golden orbit, NNs should be appropriate for obtaining it via a quantitative approach to look and adjust methods (such as operator tweaking) which use pattern recognition to address hardware and software limitations, inaccuracies, errors, and imprecise knowledge or understanding of effects such as annealing and hysteresis. Insights from NNs allow one to define feasibility conditions for LCSs in terms of design constraints and tolerances. Hardware and software implications are discussed and several LCSs of current interest are compared and contrasted