摘要
依据小波函数的非线性逼近能力和神经网络的自学习特性,提出一种适合高维输入的小波神经网络建模方法,这种网络结构类似于多层感知器,不同的是隐层神经元的激励函数为小波基函数.为使小波神经网络具有更高的学习精度和更快的收敛速度,将遗传算法、小波神经网络和梯度下降算子结合起来,提出一种遗传小波神经网络.将该网络应用于冷轧轧制力的预报,仿真结果表明预报精度大为提高.
Based on the function approximation ability of wavelet and the learning characteristic of neural network, a wavelet neural network (WNN) is introduced to handle the high dimension input problem. The structure of the WNN is similar to that of multi-layer perception, but the active function of hidden nodes is replaced by a wavelet base function. In order to obtain higher accuracy and faster speed, a wavelet neural network based on hybrid genetic algorithm (GAWNN) is put forward, which combines genetic algorithm with wavelet analysis and neural network and gradient descend operator. The application of GAWNN to cold mill rolling force prediction gives better results than typical model and force prediction precision is improved. Simulation results demonstrate the effectiveness of the methodology.
出处
《控制与决策》
EI
CSCD
北大核心
2004年第10期1129-1132,共4页
Control and Decision
基金
国家自然科学基金资助项目(60274024).
关键词
小波神经网络
混合遗传算法
轧制力预报
Computer simulation
Genetic algorithms
Multilayer neural networks
Optimization
Predictive control systems
Wavelet transforms