一类具比例时滞递归神经网络的无源性

Passivity of a class of recurrent neural networks with proportional delays

研究一类具比例时滞递归神经网络的无源性。通过非线性变换,将具比例时滞的模型等价变换成常时滞模型;通过构造合适的Lyapunov泛函和线性矩阵不等式,充分利用神经元激活函数的条件,得到两个新的保证具比例时滞递归神经网络无源性的条件。数值算例及其仿真结果验证了结论的正确性。

Passivity of a class of recurrent neural networks with proportional delays is investigated. By applying a nonlinear transformation, the primary model with proportional delays is transformed equivalent- ly into a model with constant delays. By constructing proper Lyapunov functional and linear matrix ine- quality （ LMI）, and fully using the conditions of neuron activation functions, two new passivity conditions are proposed to ensure the passivity of recurrent neural networks with proportional delays. Two numerical examples and their simulation results are given to testify the effectiveness of the proposed criteria.