Abstract: The process of water inrush disaster accompanied by the information change of rock and soil within the preventing structure changed accordingly. This study conducted an in-depth fusion analysis of multi-physical field information for potential water inrush channels. The logical symbiosis relationship of multi-information was revealed, and a theoretical method for identifying the evolution state of water inrush was established. Two typical similarity model tests of water inrush were conducted to simulate the progressive failure of rock mass and filling structure instability. Then, we proposed the correlation between the multiple monitoring information and the quantitative characterisation function. Based on the theory of principal component analysis, this study revealed the logical symbiosis relationship between multiple parameters and determined the influence weight of the multi-physical field information in the surrounding rock, which provided a theoretical basis for the monitoring design and early warning of water inrush disaster. A comprehensive method for the evolution state of water inrush disaster was established, combined with the extreme point and the stagnation point of the fitting curve representing the changing trend of the function, Finally, the evolution stages of these two types of water inrush are divided into four periods: calming period, development period, abrupt period and post-disaster period. The energy discrimination method of the occurrence probability of water inrush was put forward at any moment. The theoretical method provides a reference for monitoring and early warning of water inrush disaster in tunnels and underground engineering.

Key words: water inrush, multiple information, status identification, principal component analysis