We present the Phenomenological Reconstruction of Complex Systems (PRCS) method for plasma diagnostics, applied to bolometer data from the Large Helical Device (LHD, shot ›187366). Instead of conventional amplitude or spectral analysis, we reconstruct the phase space, compute invariants (𝐽, Φ) in a multiscale regime (𝜏 ∈ 1, 5, 10, 20), and extract stable dynamical patterns. Four patterns are identified, characterised by low transition entropy (0. 45 bit), a topological invarant 𝐿𝑘 ≈ 1. 0 for all regimes, and a characteristic frequency (1. 08 Hz for the dominant pattern). We show that the structural risk 𝑅 (𝑡) peaks precisely at pattern switchings, not within stable intervals. The method works on open data, requires no model assumptions, and can serve as a basis for phase-state diagnostics and the design of onto-control system
Sergey Aleksandrovich Mazein (Sun,) studied this question.