将误差反传 (BP)算法和遗传算法 (GA)有机地结合在一起 ,提出了一种新的算法 BP- GA。采用 BP- GA算法 ,设计了一个两层前向 L SI神经网络。作为神经网络的关键部件 ,提出的新型神经元性能优越。它的激活函数与理想sigmoid函数拟合很好 ;可实现对阈值及增益因子的编程并且不同增益因子下饱和输出电压值相同。采用标准 1.2 μmCMOS工艺的模型参数 ,对该两层前向神经网络电路进行的HSPICE模拟证明了它有解决异或 (XOR)问题的能力 The BP algorithm and the genetic algorithm (GA) are combined organically, and a new BP-GA algorithm is proposed. Using BP-GA algorithm, a two-layer forward L SI neural network is designed. As a key component of neural network, the proposed new neuron has superior performance. Its activation function fits well with the ideal sigmoid function; programming of thresholds and gain factors is achieved and the saturation output voltage values ​​are the same for different gain factors. Using the model parameters of a standard 1.2 μm CMOS process, HSPICE simulations of the two-layer forward neural network circuit demonstrate its ability to solve XOR problems