The characteristics of stability diagram of several typical flow patterns are studied, and the signals of flow patterns under 47 kinds of flowing conditions are identified with stochastic subspace parameter identification. The results show that the characteristic of a stability diagram can reflect the internal characteristics of complex time series, the characteristics of time series can be quantified by extracted straightness characteristics and the differences of characteristics of stability diagrams of three typical flow patterns. The characteristics of stability diagrams of bubble flow are the most confused and followed by Mist flows'. Slug flows' are the most regular. The extraction and recognition for characteristics of gray fluctuation sequence of gas-liquid two-phase images and the accurate classification for flow pattern samples provide a new path for the classification of multi-phase flow patterns. At the same time, the stability diagram analysis method provides a useful exploration for the further revelation of flowing mechanism of multi-phase flows.