This invention relates to the field of
big data analytics, specifically to a
big data-based
residual oil hydrotreating analysis
system, comprising a
data acquisition module, a path identification module, a disturbance confirmation module, an efficiency evaluation module, and a section output module. In this invention, multi-source
dynamic data from the
residual oil hydrotreating process is synchronized in time and reconstructed periodically.
Path deviation behavior is screened based on dual thresholds for the direction and magnitude of feed rate changes. Disturbance trends are classified and identified by combining
rate difference gradients. Furthermore, the directional matching relationship between aromatic concentration and hydrotreating efficiency is correlated, constructing a
coupling response labeling mechanism among multi-source indicators. This enables dynamic tracking of the intrinsic causal relationship between feed disturbances and efficiency changes, clearly distinguishing between trend fluctuations and random disturbances. High-confidence
coupling identification labels are output within the overlapping intervals of multi-
source data, enhancing the
perception of complex changes in the reaction process and improving the real-time performance and accuracy of operational condition judgment.