The Research Of Multi-spectral Radiation Temperature Measurement Based On Improved Genetic Algorithm

Temperature is an important parameter in the work process of gas turbine. When high performance gas turbine at run time, accurate measurement of surface temperature of hot end rotating parts is the key to correctly evaluate the cooling effect and working conditions of turbine blades, and make sure the engine safety. So it has important significance for how to achieve accurate measurement of the temperature of the turbine blade. In this paper, we use genetic algorithm to realize multi-spectral thermometry, through the analysis of experimental results we found algorithm due to parameters fixed, which can't meet the requirement of the actual parameters dynamic and change, so that the accuracy of the obtained temperature and emissivity is not high. In this paper in-depth studies on this problem, and proposed a new solution:(1)Genetic algorithm parameters are fixed, which can't meet the requirement of the actual parameters dynamic and change. To solve the problem, we use improved adaptive genetic algorithm optimization objective function. Improved adaptive genetic algorithm uses elitism thus ensuring the best individual of each generation is not destroyed, and to improve the accuracy by improving the convergence crossover and mutation operator. Experimental results show use improved adaptive genetic algorithm can improve the accuracy of the temperature and emissivity.(2) The accuracy of the obtained temperature and emissivity is not high, for this problem,this paper proposed an adaptive double population genetic algorithm. The multi-spectral temperature measurement method based on Adaptive double population genetic algorithm is in the double population genetic algorithm adaptive adjustment parameters. Double population genetic algorithm evolved two populations simultaneously. The two populations respectively independent execute the selection, crossover and mutation operations, one pay attention to local search, and the other pay attention to global search, the two populations adaptive adjustment parameters respectively, fast and accurate convergence to the optimal solution. Then at the end of each generation run select the random individuals and good individuals from the population respectively to exchanged, and the immigration operation in the algorithm can balance local and global capabilities of the algorithm, thereby improving the efficiency of the algorithm. Experimental results show adaptive double population genetic algorithm improved the multi-spectral measurement accuracy and emissivity precision.