在光伏发电系统中，经常存在电池板局部被遮挡的情况，造成电池板出现多峰极值的现象。传统的最大功率点(MPPT)搜索方法常常会陷入局部极值，从而错过或丢失系统的全局最大功率点，甚至产生振荡导致系统输出不稳定。提出一种改进的蝙蝠算法(IBA)，并应用到太阳能阵列存在局部阴影条件下的最大功率点寻优控制中。通过混沌初始化，对群体的初始位置进行更新，增加种群的均匀性和遍历性；引入自适应惯性权重，使算法在优化前期具有较强的全局搜索能力，后期有较强的局部收缩能力，同时引入Levy 飞行来产生跳跃速度，跳出局部极值；引入动态收缩区间，有效地减小算法的搜索范围。以上改进，避免了种群受到局部极值的影响而过早收敛。光伏发电系统的仿真表明：在其受到局部遮挡而出现多峰极值的情况下，改进的蝙蝠算法能够快速找到全局最优点，并且精度高。

In photovoltaic systems, the output power curve of solar battery has multiple peaks, under the partially shaded condition. Traditional maximum power point tracking (MPPT) search method often traps in local extremum, which causes the loss of the global maximum power point even generates oscillation and leads to instability of output. An improved bat algorithm (IBA) is proposed and used to find global optimal point, by introducing chaos search strategy in initial arrangement which can improve the uniformity and ergodicity. The self-adapting weight is introduced to enhance the global searching ability of previous processing and the local searching ability of late processing, and Levy flight is introduced in the same time to create the saltation velocity to jump out the local extremum. Dynamic contraction is also used to decrease the search section more effectively, so as to avoid premature convergence of the population affected by the local extremum. The simulation shows that modified bat algorithm can find the global optimal point fast, with high precision, under the partially shaded condition.