| With the rapid development of agriculture,industry,manufacturing and other industries,human’s demand for accurate temperature measurement is also getting higher and higher.As a non-contact temperature measurement method,multispectral radiation temperature measurement technology has been widely used because of its fast response speed,wide measurement range,and not affecting the temperature field of the measured target.The core of this technology is to process the output data of the instrument,and the key problem of data processing is that the spectral emissivity of the measured target is difficult to measure accurately.Spectral emissivity is a physical quantity affected by many factors,such as object material,surface roughness,degree of surface oxidation,temperature,and wavelength.Most current data processing methods require a hypothesis model of emissivity in advance,but it will inevitably bring error to the measurement.To solve this problem,the data processing algorithm for multispectral radiation temperature measurement is studied based on the optimization principle.Firstly,a two-time true temperature inversion algorithm based on the multi-segment linear model of emissivity in traditional model method is proposed,aiming at the poor generality problem of the emissivity in the traditional model method.This algorithm can constructs a multi-segment linear emissivity model according to the radiation characteristics of the measured target,so as to avoid the error of the traditional emissivity model,and then solves the problem of channel temperature inconsistency caused by the emissivity model through two inversions.The accuracy and efficiency of this algorithm are analyzed in true temperature inversion,and compared with other true temperature inversion algorithms,which proves that the proposed algorithm is effective.Secondly,aiming to the defect of multi-segment linear emissivity model itself,in order to eliminate the error of it,a true temperature inversion algorithm based on multi-objective constraint optimization is proposed.Instead of making any assumptions on the emissivity of the measured target,the algorithm directly constructs multiple objective functions and constraints based on the reference temperature model,and then solves the constraint optimization problem by the mixed penalty function method.Taking tungsten and hypothesis target as the experimental object,the algorithm is used for simulation,and the results verify the effectiveness of the algorithm.Finally,a true temperature inversion algorithm based on ridge estimation and sequence quadratic programming is proposed for the disadvantages of least square method and complicated constraints in the multi-objective constrained optimization algorithm.The algorithm not only uses the least square method but also uses the least absolute method in the construction of the objective function,its constraints are directly determined by the defined range of the emissivity,and proposes an adaptive compression factor to calculate the initial point with ridge estimation.The accuracy,efficiency and universality of the algorithm are verified by simulation experiments on hypothetical targets,tungsten and rocket nozzles. |
