中文摘要：

算法采用系统分解理论将系统约束松弛，把机组组合问题分解为2层优化问题。上层通过拉格朗日乘子的自适应调整来协调单个机组的子系统，下层采用遗传算法求解单个机组独立的子系统优化问题。对拉格朗日乘子的自适应调整明显减少了对偶间隙的振荡现象．对遗传算法中交叉变异算子自适应的调整有效地克服了早熟现象。算例表明可行解的质量高、收敛速度快，与传统算法相比具有更高的自适应性，适用于大规模、复杂系统的机组组合问题的求解。

英文摘要：

The adaptive system optimization algorithm uses the system decomposition theory to relax the system constraints and transforms the unit commitment into a two-level optimization problem. The high level optimization coordinates the subsystems of individual unit to reduce the vibration of duality gap by the adaptive adjustment using Lagrangian multiplier, while the low level optimization applies Genetic Algorithm to optimize each subsystem of individual unit. The probabilities of crossover and mutation are adaptively changed for each generation to avoid the pre- maturity. Numerical results show that,the proposed algorithm has high quality solution,fast convergence speed and good adaptability,suitable for the unit commitment of large and complex power system.