Research On Determining Algorithm Of Weather Forecast KF Parameters Based On Simulated Annealing Method

In recent years,with the rapid development of computer technology,the forecast of meteorological science is more and more accurate,and the optimization method in the field of mathematics as a suitable mathematical tool can also help the progress of meteorological science.One of the most outstanding is the application of a variety of optimization methods to improve the necessary parameters in meteorological prediction,which can make the irregular atmospheric movement more accurate due to the addition and adjustment of relevant factors.Simulated annealing algorithm belongs to Monte Carlo global optimization method in continuous data assimilation algorithm.It generates a new value by adding a temperature dependent Cauchy like disturbance to the current value,and then adopts metropolis probability acceptance criterion,which can make the current value jump out of the solution space of local minimum and reduce the possibility of falling into local minimum,In this process,the simulated annealing algorithm is controlled according to the traditional annealing plan,and finally the optimal solution is found.This method of accepting the new model makes the simulated annealing algorithm no longer a local search algorithm,but a new global optimal algorithm.In this dissertation,the traditional simulated annealing method and the improved simulated annealing method are used to adjust the five key parameters of KF scheme in the weather research and forecast model.A group of adjustable parameters are obtained and more reasonable values are provided to approximate the observed climate.Finally,the performance of traditional simulated annealing method is analyzed,a new improved simulated annealing method is obtained by improving the disturbance process and tempering temperature for many times,and its five key parameters in KF scheme are adjusted to make the final solution more reliable.In order to verify the feasibility of simulated annealing method,this dissertation first solves the problem of shortest route and best value through two examples to verify the feasibility of simulated annealing method.Then,the improved simulated annealing method is applied to KF scheme,the observation data and forecast data are compared and analyzed briefly,Experiments show that the improved parameters have better ability to predict the weather.