Clustering of paraffin-based hybrid rocket fuels combustion data

Abstract

Clustering was applied to image data of hybrid rocket combustion tests for a better understanding of the complex flow phenomena. Novel techniques such as hybrid rockets that allow for cost reductions of space transport vehicles are of high importance in space flight. However, the combustion process in hybrid rocket engines is still a matter of ongoing research and not fully understood yet. Recently, combustion tests with different paraffin-based fuels have been performed at the German Aerospace Center (DLR). For a detailed analysis, the combustion process has been captured with a high-speed video camera, which leads to a huge amount of images for each test. In the end, a large dataset with a total number of 30 000 images for each combustion test has to be analyzed. In order to catch the essential flow structures, the combustion dataset was clustered with a K-means++ algorithm. Since the algorithm might converge to local optimal solutions, expensive repetitions have been performed in order to ensure that a global

Local anomaly detection in hybrid rocket combustion tests

“…Therefore, an automatic analysis method is needed in order to get a faster and deeper insight into the combustion process. An overview of the different flow phases can be obtained by an automatic clustering of the dataset (Rüttgers et al 2020). This gives essential information on the mean burning behavior.…”

Measurement of Hybrid Rocket Solid Fuel Regression Rate for a Slab Burner using Deep Learning

“…Neural networks in general have also been recently utilized in the combustion field, including maximizing power density in proton exchange membrane fuel cells [48], creating a fuel consumption model to predict the most efficient route for truck transportation [41], and to classify solid fuels based on ash content, volatile matter, and fixed carbon [16]. Similar machine learning processes have been used in hybrid rockets to recognize the different burning phases of solid fuel in a similar 2D slab burner experiment [45]. The most relevant work to the goal of this study is the particle shape and size regression detection, an example, BubCNN, presented by [19], where neural networks were used to detect the shape and size of bubbles in a gas-liquid multiphase flow while segmenting them from the background.…”