对于求解不等式约束优化问题,将线搜索和滤子方法相结合提出了一种新的线搜索滤子序列二次规划(filterSQP)方法.该方法克服了传统的SQP方法二次子问题不相容的困难,并利用滤子避免了罚函数的使用.同时在合理条件下证明了此方法具有全局收敛性质.
The paper presents a line search filter sequential quadratic programming(SQP) method for inequality constrained optimization.Compared to traditional SQP methods,the advantages are that the quadratic programming(QP) subproblem is always consistent and the penalty function is not required by filter strategy.Under some mild conditions the global convergence can be induced.