Neural network-based study on the correlation between exhaust plume images and combustion chamber pressures of the throttleable hybrid rocket motor

The relation between exhaust plume images and combustion chamber pressures of the throttleable hybrid rocket motor has not gained much attention. A neural network method is proposed to explore the correlation between exhaust plume images and combustion chamber pressures. Based on the idea of classification, we classified the combustion chamber pressures according to a piecewise function. The image of each frame of the input video was matched with each stage of the combustion chamber pressure to establish their corresponding relation with the machine learning method. In the training process, the pressure data were used as labels to match the corresponding exhaust plume images. In the testing process, after the input of the video, the combustion chamber pressures were automatically obtained according to the images. The results show that the exhaust plume images of different combustion chamber pressures present significant differences. Besides, with the images of exhaust plume as input, the test results of the neural network method show an 86.40% accuracy in the identification of the combustion chamber pressures.