Gas-water two-phase flow process is complex and dynamic. Real-time monitoring of flow status evolution is helpful to the safe operation. The progress of multi-sensor technology makes it possible to accurately measure the flow information, which lays a data foundation for the establishment of flow status monitoring model. Due to the instability and randomness of flow, there are both nonstationary and stationary signals in the measurement data. In this paper, the augmented Dickey Fuller (ADF) test is used to distinguish nonstationary signals from stationary signals. Cointegration analysis (CA) can deal with the problem that the fluctuation of nonstationary signals masks the change trend of flow status. For stationary signals, kernel canonical variate analysis (KCVA) is used to realize the nonlinear mapping and obtain the canonical features reflecting the dynamic changes. Two monitoring indexes are built and combined to monitor the flow status. CA- KCVA method is applied to the data of gas-water two-phase flow in horizontal pipe. The monitoring results prove the effectiveness.