基于SQP算法的概率积分法参数反演

Parameter Inversion of Probability Integral Method based on SQP Algorithm

针对传统优化及智能优化算法反演概率积分法参数时,易受到初值与自身参数设置影响,难以获得最优解的问题,提出采用二次序列规划(Sequential Quadratic Programming,SQP)算法反演概率积分法函数模型中的参数。通过模拟不同误差环境下的仿真数据,分别采用模矢法、遗传算法与SQP算法进行对比实验。结果表明,SQP算法反演参数结果可靠,不受初始值设置影响。并且在不同的误差环境影响下,SQP算法对各个参数的反演精度高,大部分情况下优于遗传算法与模矢法,对随机误差与粗差具有良好的抗干扰能力。

Solving the divergence problem of parameter inversion in the probability integration method model by optimization algorithms or intelligent optimization algorithms,it is vulnerable to get optimal solution by the initial value and its own parameter settings.It is proposed to use the Sequential Quadratic Programming(SQP)algorithm to invert the parameters in the probability integral method.Using the simulation data under different error environments,the comparison experiments were performed by the modular vector method,genetic algorithm and SQP algorithm.The results show that the inversion results of SQP algorithm are not affected by the initial value setting.And under the influence of different error environments,the SQP algorithm has high inversion accuracy of each parameter,which is better than genetic algorithm and mode vector method in most cases,and has good anti-interference ability to random errors and gross errors.