文章提出一种求解机组组合问题的分支定界方法。利用热率和透视割平面将机组组合模型做近似线性化处理,将混合整数二次规划转化为混合整数线性规划进行求解。为提高分支定界的计算效率,进行上层变量固定并优先搜索机组启停状态最接近0.5的节点。数值结果表明,此法与直接求解混合整数二次规划的发电费用相当,但计算时间占优,适合求解大规模机组组合问题。This paper presents a branch and bound method for solving the unit commitment (UC) problem. The UC model is approximated to linear models by using the heat rate and perspective cutting plane, and the mixed integer quadratic programming model is transformed into mixed integer linear programming. In order to improve the computing efficiency of the branch and bound, the upper-level variables are fixed and the node closest to the value of 0.5 is searched first. Numerical results show that the generation costs by the proposed method are almost equivalent to the ones by directly solving the mixed integer quadratic programming of UC. However, the proposed method has an advantage in terms of computation time and is suitable for solving large-scale UC problems.
