A calibration method for infrared measurements on building facades based on a WOA-BP neural network

As a critical method for analyzing the thermal properties of buildings, the audit results of infrared thermography are affected by the accuracy of temperature measurement. However, it is still a challenge to determine the actual temperature of the target surface due to various factors (e.g., object emissivity, ambient temperature, relative humidity, measurement distance, etc.). To this end, this paper proposes a robust temperature calibration model by utilizing a BP neural network trained with numerous sets of laboratory temperature measuring data, using the Whale Optimization Algorithm (WOA) to optimize the initial network parameters, and rigorously validating the model with field measurements. The comparison of WOA-BP with traditional BP, SVM, RBF, PSO-BP and GWO-BP confirms its superiority in terms of correction error and robustness, which can effectively solve the problems of temperature measurement deviations in building facades.