混沌神经网络已被证明是解决组合优化问题的有效工具,但单一化的退火因子无法同时满足准确性和速度性两方面要求,因此改变传统的混沌方式以提高搜索速度和精度就变得尤为重要.文中将Sigmoid函数转化为小波函数可以有效地解决该问题,通过将Sigmoid函数转化为Mexican hat小波函数,以及引入Shannon小波和Sigmoid函数加和组成的非单调激励函数这两种方式,提高了搜索效率和准确度,并用这两种新的模型对两种优化问题进行仿真.仿真结果表明小波混沌神经网络无论在全局最优解的搜索效率还是精确度上都明显优于传统的混沌神经网络.可知将小波函数引入混沌神经网络是极具研究潜力的.%Chaotic neural network has been proved to be a valid tool for solving combinational optimization problems. But the single factor of the annealing cannot meet in the terms of both accuracy and speed requirements. So to change the traditional way to improve the search of chaotic speed and accuracy becomes more important. This Sigmoid function into the wavelet function can solve the problem, through the Sigmoid function into a Mexican hat wavelet function, and the introduction of Shannon wavelet and Sigmoid function and composition of additional non-monotonic activation function of these two methods to improve the efficiency of search and accuracy. And use these two new models to simulate two kinds of optimization problems. Simulation results show that the wavelet chaotic neural network optimal solution in the search speed and accuracy are much better than the conventional chaotic neural network. Show that the introduction of the wavelet function to chaotic neural network is a great potential.
展开▼
