Investigation of a Data-Driven Method for Analyzing Impulse Tunnel Aerothermal Test Data

Abstract

The test time of impulse tunnels is short, and the heat flux transducers can be easily interfered in the test time. Thus the aerothermal test data analysis of impulse tunnels is complex and usually carried out manually, which leads to low efficiency and nonuniform standards. In this paper, the signals of heat transducers used in the 1 m gun/shock dual-mode impulse tunnel at the China Academy of Aerospace Aerodynamics were investigated. A series of key parameters, e.g., the FFT voltage profile, the correlation profile of nozzle total pressure and transducer heat flux, as well as the time derivative of the heat flux profile, were determined. The characterization capabilities of these parameters with respect to significant features, e.g., the start/end time of the nozzle flow, the time boundary of different operation modes and the effectiveness of the transducer profiles, were certified. Moreover, a profile effectiveness estimation model, a test time interval selection algorithm and a transducer valid time selection algorithm were developed based on convolutional neural networks and density-based clustering algorithms. Thereby the aerothermal test data of impulse tunnels can be analyzed automatically with these data-driven methods. The results show that the accuracy of the profile effectiveness estimation model is as high as 98%. According to the comparison between algorithm and manual results in dealing with 3 568 profiles, about 91% of the results show a difference of less than 10%, which implies that the data-driven method has the value of application to wind tunnel tests.

Keywords

• impulse tunnel
• aerothermal test
• heat flux
• convolutional neural networks
• density-based clustering