【作者】 王伟； 王书宁； 高林；

【Author】 WANG Wei WANG Shuning GAO Lin (Department of Automation, Tsinghua University, Beijing,100084)

【机构】 清华大学自动化系系统工程所；

【摘要】 凸可分分片线性目标函数下的优化问题得到了广泛的研究,但是任意的连续分片线性函数的优化问题尚没有得到较好的解决。本文通过引入任意的连续分片线性函数的一般紧凑表达形式minimax 方式,将分片线性函数优化问题转化为线性规划问题,从而获得优化问题的全局最优解。对于任意的连续函数优化问题,通过使用minimax表达模型的分片线性函数逼近连续函数,可望获得连续函数优化问题的近似解。更多还原

【Abstract】 Piecewise linear function optimization is important in theoretical and practical application. The available methods in the literature have developed to minimize continuous piecewise linear objective functions which are separable and convex. Concerned on arbitrary continuous piecewise linear functions, this paper presents an algorithm to transform the optimization problem to linear programming. By canonical representation of piecewise linear functions named minimax model, the global optimal solution will be gotten.更多还原