针对复数满秩矩阵的Moore-Penrose逆问题,采用一种新型的递归神经网络(ZNN)进行求解.构造3个不同的复数矩阵误差函数,利用ZNN设计公式推导得到对应的不同复数ZNN模型.为了便于计算机仿真,采用向量化技术将所得到的ZNN模型由矩阵形式转换为矩阵向量形式.计算机仿真结果表明了所得到的3个复数ZNN模型在求解复数满秩矩阵Moore-Penrose逆时的可行性与有效性.
For the solution of complex full-rank Moore-Penrose inverse,a new kind of recurrent neural network(i.e.,ZNN)is adopted.First,three different complex error functions are constructed.Then,by employing the ZNN design formula,three different complex ZNN models are obtained.Furthermore,in order to facilitate computer simulation,the complex ZNN models are transferred from the matrix form to the matrix-vector form based on the vectorization technique.Finally,computer-simulation results prove the feasibility and effectiveness of the three proposed complex ZNN models when solving the Moore-Penrose inverse of complex full-rank matrix.