关键词: 液体推进剂火箭发动机;非线性方程;回归分析;计算数学

Abstract: Due to the shortcomings of conventional regression algorithm based on MQE(mean squared error)that can not be easily overcomed,a hybrid evolutionary regression algorithm to this problem presented and the structure and coefficients of the nonlinear regression equations can be regressed with this algorithm simultaneously By setting the maximal number limit of nonlinear items for the regression equation,this hybrid algorithm is firstly used to obtain some initial seeded regression equations,which basically meet the requirements on the regression accuracy and the number limit of nonlinear items,then these initial seeded equations can be simplified and optimized progressively through Evolution Strategies to form a parsimonious nonlinear regression equation,this simplification and optimization process is under the supervision of a criterion:to minimize the residual and the number of nonlinear items.Different cross over and mutation operators are specially designed for the Evolution Strategies to make the hybrid nonlinear regression of structure and coefficients possible and efficient In addition,several kinds of statistieal nonlinear relationship in liquid rocket engine are regressed based on experimental data samples,the residual distributions show that this hybrid evolutionary nonlinear regression algorithm is effective and pragmatic.

Key words: Liguid propellant rocket engine;Nonlinear equation;Regression analysis computational mathematics