【摘要】 将混沌扰动直接添加到Hopfield网络中以提高网络在函数优化计算中的寻优能力.在寻优过程中,通过不断衰减混沌扰动幅度及混沌扰动的接受概率来实现混沌的模拟退火.接收概率衰减速度的调节可以控制混沌退火的速度,从而影响网络的收敛速度.网络在优化过程中经历了混沌粗搜索和梯度下降精搜索两个阶段.利用混沌的随机性和遍历性等特点,网络可以到达全局最优点附近,最终获得全局最优解.仿真结果证明了该方法的有效性.更多还原

【Abstract】 The performance of Hopfield neural networks was improved through the addition of chaos noise. During the process of optimization, chaos simulated annealing was realized by decaying the amplitude of the chaos noise and the probability of accepting continuously. Adjusting the probability of acceptance could control the speed of chaos simulated annealing, and influence the rate of convergence. The process of optimization was divided into two phases: the coarse search based on chaos and the elaborate search based on gradients. Utilizing the randomicity and ergodicity property of the chaos, the network can get around the global optimal points, and obtain the solutions of optimization. Simulation results proved the validity of the algorithm.更多还原