Frequent sandstorms and severe dust accumulation in desert environments make photovoltaic arrays highly susceptible to shading. In addition, large temperature fluctuations between day and night and intense ultraviolet radiation can accelerate module failures and degrade optical performance. These adverse conditions inevitably result in a reduction of power generation and a decline in the output performance of solar arrays. Meanwhile, each photovoltaic array in desert power plants is large in scale, and reconfiguring such arrays involves a high computational burden. These factors demand high optimization capability and fast convergence from reconfiguration algorithms. To address these challenges, this study develops a PV array reconfiguration approach using the Improved Multi-Objective Black Kite Algorithm (IMOBKA). The method aims to simultaneously maximize the output power of the array, minimize the standard deviation of row currents, and reduce the range of row currents, adopting a 3-opt exchange strategy to further enhance the black kite algorithm's search capability and accelerate convergence. The proposed method was simulated on a 9 & times; 9 photovoltaic array under six shading conditions, including four types of shading specific to desert environments and two compound shading conditions combining module optical degradation and faults. The reconfiguration performance of IMOBKA was compared with five existing methods, namely TCT, Modified Sudoku, POA, BKA, and MOGWO, using three evaluation metrics: fill factor, power loss percentage, and power enhancement percentage. The results show that under six different scenarios, the output power of the PV array reconfigured by IMOBKA is improved compared to other methods. Compared to the TCT configuration, the power enhancement percentages are 12.13%, 10.47%, 18.54%, 17.46%, 9.57%, and 12.27%, respectively. It is evident that IMOBKA demonstrates good performance in addressing module failures, optical performance degradation, and shading issues.