Study On Optimisation Of Inversion Algorithms Based On Multispectral Radiometric Methods

Multispectral radiometric thermometry is a non-contact method of temperature measurement that is used in a wide range of industrial applications.The method uses spectral radiation information of several wavelengths to invert the true temperature of the measured object(true temperature)by means of a corresponding algorithm.However,due to the combined effects of wavelength,temperature,material composition,and surface condition,it is often difficult to obtain accurate information on the spectral emissivity of the object being measured,which affects the accuracy of the true temperature inversion.Therefore,the optimization of the inversion algorithm for multispectral radiometric temperature measurement is of great importance.To address the above issues,the specific research in this paper is as follows:Firstly,to address the problem of poor generalisability of the assumed emissivity model,a reference temperature model is used to transform the true temperature solution problem into a non-linear optimization problem and to construct a non-linear optimization objective function for multispectral radiometric temperature measurement,thus enabling the inverse performance of the true temperature of the measured object without a priori data.Secondly,the SA-BFGS true temperature inversion algorithm is proposed to address the problem that the initial value of the BFGS algorithm can easily fall into a local optimum if it is not chosen properly.This algorithm firstly uses the Simulated Annealing(SA)algorithm to find the global optimum of the objective function,then introduces the Wolfe line search method to determine the optimal step size of the BFGS(Broyden-Fletcher-Goldfarb-Shanno)algorithm,and finally uses the BFGS algorithm to solve iteratively to obtain the true temperature of the measured the real temperature of the measured object was obtained by iterative solution of the BFGS algorithm.The simulation experiments of six materials and tungsten metal show that the relative error of the improved SA-BFGS algorithm in inversion of true temperature is less than 1.5%,and the average inversion time is less than 1 second,verifying the effectiveness of the improved SA-BFGS algorithm proposed in this paper.Finally,to further improve the computational speed and accuracy of the SA-BFGS algorithm,the IMP-SA-BFGS true temperature inversion algorithm is proposed.The algorithm uses a discrete stochastic optimized thermostatic simulated annealing algorithm to perform global optimization of the objective function and improve the algorithm inversion speed.A modified weak Wolfe-Powell(MWWP)line search method is used to obtain the optimal step size and improve the accuracy of the algorithm.The simulation experiments of six materials and tungsten metal show that the relative error of the improved SA-BFGS algorithm in inversion of true temperature is less than 1.5%,verifying the effectiveness of the improved SA-BFGS algorithm proposed in this paper.