Overview: Gas-liquid two-phase flow is widely used in the industrial field, and one of the typical flow patterns is bubble flow. Therefore, measuring the bubble flow phase distribution parameters is of great significance for studying the characteristics of two-phase flow and industrial production. Optical technology is widely used in small-channel gas-liquid two-phase flow detection, and it can be divided into invasive and non-invasive. Invasive optical detection methods will affect the characteristics of gas-liquid two-phase flow to a certain extent, so non-invasive measurement is an important research direction of optical technology. Some scholars have proposed a method of measuring phase distribution based on the characteristics of single-wavelength light intensity distribution, but the established model is only suitable for the horizontal pipeline, and the resulting error is large. To solve this problem, a dual-wavelength measurement model of bubble flow phase distribution parameters is proposed in this paper, to obtain more optical signal characteristics of laser passing through bubble flow in small channels and reduce the measurement error of phase distribution parameters. The phase distribution and parameter distribution of vertically rising gas-liquid two-phase bubble flow in a small channel are measured and studied by the dual-wavelength transmission method. The light intensity distribution of dual-wavelength laser passing through gas-liquid two-phase flow is calculated based on the principle of the geometric optics, and then the characteristic quantity of dual-wavelength light intensity distribution is extracted. An identification model of bubble flow phase distribution parameters based on the dual-wavelength measurement theory is established. Trace Pro is used to trace the refraction trajectories of 445 nm and 635 nm laser passing through the pipe section containing bubbles with different phase distribution parameters, and the corresponding light intensity distribution curves are obtained. The key features of the corresponding double wavelength light intensity distribution curves are studied, and three kinds of characteristic parameters are extracted: missing part length, missing part offset, and double wavelength interval length. The neural network is trained by using the simulated feature data set, and the trained neural network is used to predict the phase distribution parameters of bubble flow in the experiment. The simulation results show that the prediction results are in good agreement with the simulation data, and the relative error is within ± 5%. The average absolute errors of the established model for the prediction of bubble center position and radius are 0.018 mm and 0.007 mm respectively. The results show that the dual-wavelength method has higher accuracy and is more suitable for the measurement of gas-liquid two-phase flow. Finally, a dual-wavelength gas-liquid two-phase flow measurement system is built. The bubble flow in the small vertical rising channel with the small gas flow is measured by using a dual-wavelength laser light source. The bubble flow is obtained by injecting gas into the stagnant liquid column. The experimental research is carried out, and the bubble sizes distribution curve is counted. In general, the dual-wavelength measurement method provides good results and is an alternative method for the measurement of bubble phase distribution parameters. This technology is helpful to measure and monitor bubble flow parameters online.Aiming at the deficiency of the existing single-wavelength method in measuring the phase distribution parameters of the vertically rising gas-liquid two-phase bubble flow in a small channel, a dual-wavelength method is proposed. Based on the principle of geometric optics, the light intensity distribution of dual-wavelength laser passing through gas-liquid-two-phase-flow is calculated, and the characteristics of light intensity distribution of dual-wavelength laser are extracted. An identification model of bubble flow phase distribution parameters based on the dual-wavelength measurement theory is established. Trace Pro is used to simulate 445 nm and 635 nm laser passing through bubbles with different phase distribution parameters, and then the features of the dual-wavelength light intensity distribution curves can be extracted. The characteristic quantity data set of simulation is used to train the neural network. The trained neural network is used to predict the phase distribution parameters of bubble flow in the experiment. The simulation results show that the average absolute errors of the model for predicting the bubble center position and radius are 0.018 mm and 0.007 mm respectively, which are better than the single-wavelength method, which proves the effectiveness and accuracy of the model. The bubble flow was measured on the experimental platform, and the three-dimensional diagram of bubble flow was reconstructed.