摘要
本文针对传统BP神经网络存在收敛速度慢、计算量大、易收敛于局部最小点的缺点,引入遗传算法与LM算法优化BP神经网络,并对优化后的BP神经网络进行训练与预测。实验结果表明:优化后的模型具有训练速度快、预测精度高的特点,更适用于大坝的实时预报。
Traditional BP neural network has disadvantages of slow convergence speed,large computing capacity and easy converge on local minimum point in the paper.Genetic algorithms and LM algorithm optimization BP neural network are introduced.The optimized BP neural network is trained and predicted.Experimental results show that the optimized model is characterized by rapid training speed and high forecast accuracy,which is more suitable for real-time forecasting of the dam.
出处
《中国水能及电气化》
2014年第5期54-58,共5页
China Water Power & Electrification
关键词
遗传算法
BP神经网络
LM算法
变形监测
genetic algorithm
BP neural network
LM algorithm
deformation monitoring
