针对BP 算法易陷入局部极小、收敛速度慢等缺点，遗传算法是全局优化算法和具有很强的全局搜索能力，遗传算法优化BP 神经网络初始连接权值和阈值形成混合算法。以安徽宣城市为例,将汛期降水量作为预测对象，前期74 项大气环流特征量、500 hPa、100 hPa 月平均高度场、月平均海平面气压场和月平均海温场资料中选取预测因子，建立汛期降水短期气候预测模型。-BP algorithm is easy to fall into local minimum, slow convergence, genetic algorithm is a global optimization algorithm and has a strong global search ability, BP neural network genetic algorithm to optimize the initial connection weights and thresholds to form a hybrid algorithm. Rainfall in Xuancheng City in Anhui, for example, as the forecast object, early 74 atmospheric circulation characteristics, 500 hPa, 100 hPa month average height field, the monthly mean sea level pressure and monthly mean sea surface temperature data, select predictor , establish flood season precipitation short-term climate prediction model.