Research On Efficient Parallel Algorithm Of Inverse Heat Conduction Problem

The Temperature Sensitivity of Parameter (TSP) is an important science research project in the invert heat conduct research to find new efficient parallel algorithms. And for develop efficient parallel algorithm is the depth of the authors’ research group for many years.This research describes the progress of the temperature sensitivity, studies non-uniform grid virtual boundary prediction on parallel algorithm, parallel computing based on two-level non-uniform grid VBF parallel algorithm. The Thermal parameters sensitivity analysis of initial boundary value problems sensitive initial boundary value problem is taking into account, Considered the expression of multiple parameters corresponding to the parallel computing method of sensitivity initial boundary value problems, the calculation results show that:for different parameters, calculation of sensitivity values are not the same. The efficient parallel algorithm of temperature parameter sensitivity is build. Finally,this article used8node processors for parallel computing, and the efficiency of parallel computing is85%.The main innovation points of this dissertation are:(1) The parallel algorithm with virtual boundary forecast based on two-grid non-uniform grids.On the double grid, the different steps of the non-uniform grid are given. On the region of virtual boundary value, establish forecast method of boundary function, the results of this algorithm show higher efficiency than the method of solving the direct iterative method that used to solve the sub-region, the computing time is about direct iterative method is about2/3.(2) Inverse heat conduction problem is used to establish the initial boundary value problem of temperature sensitivity parameter,and to cstablish the parallel computing model. According to the temperature on the sensitivity definition, establishes the initial boundary value problem of sensitivity, constructs the region splitting parallel computing model, for single and multiple parameters inverse heat conduction problem parameter sensitivity were calculated, be sensitive with the change of time and space data.(3) The fast domain splitting varied step parallel algorithm based on temperature parameter sensitivity data. According to the data of the sensitivity for different parameters, this research find the temperature parameter sensitivity is different. Therefore, a different step length parallel algorithm is obtained based on the polynomial coefficients.