What is it about?

The real-time temperature acquisition of scramjet combustors faces the challenge of high temperature and strong oxidizing environment. We use non-contact optical emission spectrum temperature measurement technology to solve the problem of real-time temperature measurement under such conditions. We tested this method in the experiment and compared it with two-wire temperature measurement and TDLAS. This method does not require a laser source and is highly resistant to noise.

Featured Image

Why is it important?

Our temperature measurement method has low equipment dependence and high noise resistance. The optical emission spectrum in scramjet engines has a low signal-to-noise ratio. So we use a convolutional neural network to match the emission spectrum of the hydroxyl group under low signal-to-noise ratio conditions. It is found that such method can effectively suppress noise and obtain temperature. In addition, because this method uses optical emission spectroscopy, it does not require an external light source and has low equipment dependence.


This article introduces a method of using interdisciplinary technology to solve temperature measurement. We position this method as a universal solution that can solve online measurement problems, and expect it to be combined with hardware to become the basic algorithm for new sensors. Although the specific scheme proposed in the article is relatively rudimentary, we believe that the scheme is not aimed at hydroxyl radicals or emission spectroscopy technology, but explains the expected vision of convolutional networks applied to new sensors.

Wanqian Xu
Harbin Institute of Technology

Read the Original

This page is a summary of: Noise-Suppressed Temperature Measurement Based on Machine Learning in a Scramjet Combustor, AIAA Journal, June 2021, American Institute of Aeronautics and Astronautics (AIAA), DOI: 10.2514/1.j060532.
You can read the full text:



The following have contributed to this page