Research On Multi-spectral Temperature Measurement Based On Improved Multi-objective Optimization Algorithm

Temperature is an important parameter.It can reflect the state of matter,especially in engineering application.Accurately measuring the temperature of a rotating turbine blade in an aeroengine is very important.It can serve to monitor the working state of the engine.Multi-spectral radiation temperature measurement is the most effective method in the area of non-contact measurement application.In multispectral calculation,the emissivity vector is put in the equation as the unknown parameter in order to reduce the dependence of emissivity information.However,for each of radiation temperature measured by each wavelength,we need to use multi-objective optimization algorithm to get the true temperature,aiming to obtain the temperature of tubine blade more accurate and stable.Since in this article the author has done the following work on this question:(1)Through the simulation of the GA algorithm and the NSGA-II algorithm,we cacluated the real temperature at 1000?,further compare these two algorithms in the aspects of accuracy?stability and calculation speed?Besides,the author passed the three or four six wavelength simulation experiments to deeply analyze of the number of wavelengths on the stability of the solution.Finally we find that the GA has a faster calculation speed and the NSGA-II has better stability in calculation.(2)In order to further improve the computational performance of traditional GA,we introduce two genetic populations,we also add the adjusting parameters and mutation factor.They can improve the population characteristics of the algorithm effectively.On the other hand,we also improve the original NSGA-? algorithm,in the crossover operator,elite strategy and adaptive crossover mutation probability.The population characteristics can be enriched as much as possible by the improvement.Besides,the differential evolution algorithm was introduced in the application of Tubine Blade temperature measurement.Finally,by the comparision of these algorithms under different emissivity models,we find that the improved GA algorithm can improve the last generation of population characteristics and the improved NSGA-II calculates the true temperature more accurately and stably for various emissivity models.But this improved algorithm is at the cost of time and since it is more suitable for offline data analysis in the application of Multi-spectral Temperature Measurement.(3)At last,we use Labview and Matlab mixed programming Simulating real-time data acquisition and calculation integration process.It highlights that the computational speed of the Differential Evolution Algorithm is higher than the original Genetic Algorithm,which is expected to be used in the real-time Multi-spectral radiation temperature measurement of Turbine blades.Finally,we use the four wavelength experimental data to further verify the computational performance of various algorithms.