采用L-M优化算法用以训练BP神经网络，使其能够拟合某一附加有白噪声的正弦样本数据。设计一个三层BP神经网络，其网络的隐含层神经元的激励函数为双曲正切型，输出层各神经元的激励函数为线性函数，隐含层有6个神经元-LM optimization algorithm used to train BP neural network to enable it to fit a white noise, sine additional sample data. Design a three BP neural network, its network of hidden layer neuron activation function of hyperbolic tangent type for the output layer activation function of each neuron is a linear function of the hidden layer has 6 neurons