Multispectral temperature measurement method based on neural network

Abstract

The invention relates to a multispectral temperature measurement method based on a neural network, and belongs to the field of spectral temperature measurement and non-contact temperature measurement. According to the invention, the spectrum is discretized, and the temperature radiation of an object is decomposed into discretized spectrums of different spectrum segments; spectral radiation of different objects at different temperatures is measured, and the spectral distribution condition is recorded through an imaging surface; the spectrum data acquired by the multiple channels is normalized; a neural network model is constructed, the preprocessed sample data is input into a neural network for training, and a nonlinear mapping relation between the spectral information and the object temperature is established; and a test sample is input, and the temperature corresponding to the test sample is obtained according to the nonlinear mapping relation between the spectral intensity information of the test sample and the object temperature. The nonlinear mapping relation between the radiation quantity and the true temperature of the target object is effectively solved through the neural network, and the method is suitable for measuring the target true temperature and the spectral emissivity of most engineering materials.

Description

technical field [0001] The invention belongs to the fields of spectral temperature measurement and non-contact temperature measurement, and in particular relates to a neural network-based multispectral temperature measurement method. Background technique [0002] The multi-spectral radiation thermometry method uses the radiation information of multiple spectra of the target object, and obtains the true temperature and spectral emissivity of the material of the target object through data processing. The measurement of true temperature depends on the spectral emissivity of the target object, and the spectral emissivity of the target is related to various factors such as material composition, surface state, measurement angle, and wavelength. To measure the true temperature of an object, it is necessary to assume a functional relationship between emissivity and wavelength, but the existing hypothetical models of emissivity and wavelength are all fixed models, and each hypothetic...

AI Technical Summary

Problems solved by technology

[0004] The technical problem to be solved by the present invention is how to provide a neural network-based multi-spectral temperature measurement method to solve the problem of the nonlinear mapping relationship between the radiation amount of the target object and the true temperature

Images