A Multi-Band Temperature Measurement Data Retrieval Method Based on the Chaotic Artificial Hummingbird Algorithm (CAHA)

To address the challenges of data processing caused by uncertain emissivity in multispectral radiation thermometry, this paper proposes a temperature retrieval method based on the Chaotic Artificial Hummingbird Algorithm (CAHA). Without relying on an assumed emissivity model, the method can automatically identify the emissivity distribution and selects the optimal output through multiple iterations to enhance accuracy. Simulations and offline tests conducted on rocket nozzles demonstrate that CAHA maintains high accuracy both in noise-free conditions and under 5% noise, with a single execution time of approximately 0.15 s. Furthermore, the method is validated through experiments on blackbody sources and candle flames: the relative error in retrieved temperature for blackbody sources remains below 0.93%, while the retrieved outer flame temperature of candle flames shows a relative error of 0.66% compared with thermocouple measurements. Combining high precision with rapid computation, this method is suitable for practical applications in radiation thermometry.