Chebyshev Chaotic Neural Networks And Their Applications In Communication

In the 1980s,hopfield neural networks(HNN)and backpropagation(BP)neural networks were proposed,which were great milestones in the history of neural network development and laid a solid foundation for the research and development of neural networks.However,due to the gradient descent strategy used by the Hopfield neural network in the optimization process,its application to larger-scale problems has been limited.In order to improve the network optimization performance,scholars have introduced chaotic dynamics,hysteresis dynamics,and wavelet techniques into the Hopfield neural network,all of which have improved the optimization performance of the network on a certain small scale.In order to improve the application of chaotic neural networks on a large scale,on the basis of existing research,a new chaotic neural network model,SCF transient chaotic neural network model,is proposed by effectively using the characteristics of chaotic dynamics.At the same time,the SCF chaotic neural network model based on the adaptive simulation annealing strategy and the SCF chaotic neural network model with perturbation are constructed,and the constructed network model is applied to function optimization,traveling salesman problem(TSP)and direct sequence-code division multiple access(DS-CDMA).Improves the relevance of the network model.The specific work of this article is as follows:(1)Through the study of Chebyshev polynomials,its higher degree of nonlinearity is used as an activation function for the newly constructed Chebyshev chaotic neurons.Control the chaotic state of neurons by adjusting the control parameters of the activation function.Through Matlab simulation experiments on the Temporal Evolution Graph and inverted fork diagram of neurons,it is verified that the newly constructed chaotic neurons have rich chaotic dynamic properties.The Chebyshev Chaotic Neural Network Model(SCF)is constructed,and the rationality of the additional energy of the chaotic neural network is verified through the unified framework theory.The network model is applied to function optimization and combinatorial optimization,which verifies the optimization performance of the model.(2)The SCF chaotic neural network model based on adaptive simulation annealing and the SCF chaotic neural network model with perturbation are constructed,and the dynamic characteristics of the model are analyzed,and the constructed model has good optimization ability through function optimization and combinatorial optimization problems.(3)The SCF model,SCF-SSA model and SCF chaotic neural network with perturbation are implemented separately to implement multi-user detectors,and the performance of the model multi-user detector is verified by simulation experiments.