Nonlinear Effects in AI and NPPs: Experimental Validation of Frequency Conversion and Self-Oscillations in Reactor Hydraulics

This paper confirms the predictions of Nobel laureates Kapitsa, Wigner, and Hopfield regarding the limitations of theoretical modeling and artificial intelligence in complex systems. Hopfield foresaw the emergence of AI systems potentially surpassing human intelligence, while Wigner and Kapitsa argued that correct solutions stem from experimental validation rather than models built on questionable assumptions. The study demonstrates that neglecting nonlinear effects, such as self-oscillations, combination, and multiple frequencies, leads to errors in AI-based machine learning following Hopfield's approach. These phenomena are identified in both the human circulatory system and the hydraulic circuit of a nuclear power plant (NPP). Experimental and computational-theoretical methods for analyzing NPP hydraulic systems are developed and validated at a VVER-type reactor. The results highlight the crucial role of frequency conversion in nonlinear hydraulic elements, where new frequencies emerge beyond the original excitation spectrum. The findings confirm that only experimental verification can provide reliable solutions in complex engineering systems.