以矩阵理论为工具，提出了三维输入向量的ＣＭＡＣ 神经网络连接权矩阵的直接计算方法．应用该方法时不必迭代求解，经直接计算便可获得连接权矩阵，所确定的连接权矩阵可使ＣＭＡＣ神经网络实现对样本的记忆 Taking the matrix theory as a tool, a direct calculation method of the CMAC neural network connection weight matrix of three-dimensional input vectors is proposed. When this method is applied, it is not necessary to solve iteratively. The matrix of connection weights can be obtained by direct calculation. The determined connection matrix can make the CMAC neural network realize the memory of the samples