论文部分内容阅读
为了克服前向神经网络的固有缺陷,提出了基于采样数据建立的含单隐层神经元的模糊前向神经网络。该网络模型利用权值直接确定法得到了最优权值,网络结构可以随采样数据的多少,自主设定隐层神经元,完成了近似插值与精确插值的转换。计算机数值仿真实验表明,模糊前向神经网络具有逼近精度高、网络结构可调和实时性高的优点,并且可以实现预测和去噪。
In order to overcome the inherent defects of the forward neural network, a fuzzy forward neural network with single hidden layer neurons based on the sampled data is proposed. The network model obtains the optimal weight by using the method of direct weight determination. The network structure can set the hidden layer neurons independently with the sampling data, completing the conversion between approximate interpolation and precise interpolation. Computer numerical simulation results show that the fuzzy forward neural network has the advantages of high approximation accuracy, adjustable network structure and high real-time performance, and can achieve prediction and denoising.