Research On The Optimization Of Load Distribution Of HSM Based On Improved GA

Set-up calculation of HSM is the basic work of the continuous rolling operation and the most important problem of it is how to make the value of each stand more feasible and make the really values of each stand's thickness sure, that is press load distribution of each stand. The reasonable load distribution can make the control of modern rolling machine implement easily, consequently get a better control results. The study on load distribution of HSM has very important realistic meanings.Flatness and gauge are two important primary criteria in strip process, and recently they are at the front line of current study on strip rolling technology. The reasonable load distribution can make the control of flatness and gauge implement easily. In this paper, the author set the hot finishing stands as the study target. In order to improve the precision of strip shape and gauge, the author carried out study on load distribution and set-up calculation on the basis of conventional mathematic models with improved Genetic Algorithms. The detail works are as follows:The author introduced the development conditions about load distribution of HSM and mainly mathematic models of set-up calculation.The author analysed and studied the Standard Genetic Algorithms (GA), improved the algorithm based on SGA named Adaptive Niche Genetic Algorithms (ANGA). Added some new improved steps, ANGA efficiently solves the problem of global convergence and improves the convergence speed. The new steps are new crossover operator, adaptive operator, calculation of niche and the prediction and dispose on immaturity. The global convergence and the convergence speed of ANGA are more satisfier than SGA.A new improved Genetic Algorithms (ANGA) of load distribution for finishing stands of HSM based on experiential load distribution as to the use and rationality of traditional load distribution methods. In the paper the author carried on an emulation experiment and compared the results optimized by ANGA with the results of experiential load distribution. The results show that the results...