<p>In the process of high-efficiency deamination, the Jet Impingement-Negative Pressure Reactor (JI-NPR) exhibits inherently multiscale and chaotic turbulence, where the underlying nonlinear dynamic mechanisms remain unclear. This study introduced a wavelet based multiscale phase-space reconstruction framework to quantitatively characterize the nonlinear dynamics of pressure fluctuations within both the low-velocity and negative-pressure separation regions under various top-negative-pressure conditions. The optimal embedding parameters (time delay and embedding dimension) were determined using the mutual information method and Cao’s algorithm to ensure accurate attractor reconstruction. Wavelet-based decomposition was subsequently integrated to reveal the scale-dependent evolution of attractor morphology and correlation dimension, enabling identification of localized stability transitions across scales. The results revealed distinct region and scale-dependent features of the reconstructed attractors and correlation dimensions. An increase in top negative pressure suppresses local chaotic intensity, thereby improving system predictability. The correlation dimension showed clear fluctuations at microscopic scales, but stabilized at macroscopic levels, indicating that top-negative-pressure regulation primarily affects micro-scale nonlinear dynamics. These findings demonstrate that the proposed multiscale phase-space framework effectively captures hierarchical dynamic organization and offers a physical interpretation of stability enhancement through pressure regulation. Furthermore, the approach provides a generalizable methodology for diagnosing nonlinear flow structures and could serve as a potential tool for energy-efficient operation and intelligent control of negative-pressure reactors. To bridge the dynamics characterization with practical performance, future work will incorporate direct energy consumption validation and multifractal analysis to extend the framework’s predictive capability.</p>