Gao Shan, Yang Qing, Liu Hailong, Han Yue, Ma Yitong, Chen Liwei, Cui Ying, Wang Tong, Zhang Zezhan, Jiang Jing, Niu Yi, Wang Chao
College of Information and Communication Engineering, Harbin Engineering University, Harbin, China.
Heilongjiang Collaborative Innovation Center of Advanced Intelligent Perception Technology, Harbin Engineering University, Harbin, China.
Rev Sci Instrum. 2025 Jul 1;96(7). doi: 10.1063/5.0268451.
Multi-spectral radiation thermometry is extensively applied in high-temperature fields. Currently, multi-spectral radiation thermometry technology still faces issues such as inaccurate identification of target emissivity models due to high-temperature background reflection and significant temperature measurement errors. This paper proposes a multi-spectral radiation thermometry method based on a Mixed Genetic Algorithm and Ant Colony Optimization Random Forest (MGACO-RFR) for emissivity model identification. By combining high-temperature background radiation with different emissivity models, an MGACO-RFR classifier is established to identify the target emissivity model. The method achieves an accuracy rate of 97.1% without noise and 94.9% with 10% added noise. After the emissivity model is obtained, the black-winged kite optimization algorithm is used to solve the radiation thermometry equation considering high-temperature background targets, thereby deducing the target temperature. The experimental results of radiation thermometry under high-temperature backgrounds indicate that for three types of samples heated in a high-temperature furnace set at 1073, 1123, 1173, and 1223 K, a cooling device is used to create a temperature difference between the samples and their environment. Applying the aforementioned algorithms to GH3044(a high-temperature alloy model 3044), GH3128(a high-temperature alloy model 3128), and thermal barrier coating sample(a sample made of material similar to that of an engine turbine blade) results in average temperature measurement errors of 3.0, 3.5, and 3.1 K, respectively. This is of significant importance for the high-precision temperature inversion of common high-temperature alloys and engine turbine blades in industrial settings under high-temperature backgrounds.
多光谱辐射测温技术在高温领域有着广泛的应用。目前，多光谱辐射测温技术仍面临一些问题，如由于高温背景反射导致目标发射率模型识别不准确以及温度测量误差较大等。本文提出了一种基于混合遗传算法和蚁群优化随机森林（MGACO - RFR）的多光谱辐射测温方法用于发射率模型识别。通过将高温背景辐射与不同发射率模型相结合，建立了MGACO - RFR分类器来识别目标发射率模型。该方法在无噪声情况下准确率达到97.1%，在添加10%噪声时准确率为94.9%。在获得发射率模型后，利用黑翅鸢优化算法求解考虑高温背景目标的辐射测温方程，从而推导目标温度。高温背景下辐射测温的实验结果表明，对于在设定温度为1073、1123、1173和1223 K的高温炉中加热的三种样品，使用冷却装置在样品及其环境之间产生温度差。将上述算法应用于GH3044（一种3044高温合金型号）、GH3128（一种3128高温合金型号）和热障涂层样品（一种由类似于发动机涡轮叶片材料制成的样品）时，平均温度测量误差分别为3.0、3.5和3.1 K。这对于工业环境中高温背景下普通高温合金和发动机涡轮叶片的高精度温度反演具有重要意义。
