论文部分内容阅读
研究了一种具有混沌特性的神经网络 ,该网络具有瞬态混沌响应 ,类似于Hopfield网络的结构 ,但有比Hopfield网络更加丰富的动力学特征、更强的全局搜索能力。通过把混沌动力学与收敛动力学相结合 ,使网络逐渐由混沌神经网络向Hopfield网络过渡 ,达到控制混沌的目的 ,并且提供一个在全局最优解附近的初值 ,有效地解决了Hopfield网络的局部极值问题。该网络模型可以用来解决复杂的非线性优化问题。
A neural network with chaotic characteristics is studied. The network has the characteristics of transient chaotic response, similar to the Hopfield network structure, but it has more abundant dynamic characteristics than Hopfield network and stronger global search ability. Through the combination of chaos dynamics and convergence dynamics, the network gradually transition from chaotic neural network to Hopfield network to achieve the purpose of controlling chaos, and provide an initial value in the vicinity of the global optimal solution, effectively solving the Hopfield network Local extreme problem. The network model can be used to solve complex nonlinear optimization problems.