The Research On Noise Chaotic Neural Network Optimization Mechanism

Hopfield neural network has been applied in broad field. However, a simpleartificial neural network is only the simplication, abstract and analog of thebiological neural network in order to realize the imitation of human brainstructure.Therefore，this kind of imitation for now is at a lower level and scientistshave been trying to explore new theory method to simulate the human brain function.In recent years, scientists find there is chaos phenomena in the brain, so bring thechaos into the Hopfield neural network has become hot research hot spots.The introduction of the chaos result the neural network has more complexdynamic characteristics and many researchers go into the research of the chaoticneural network. They have put forward many chaotic neural network model. Thispaper analysis the shortage of transient chaotic neural network and their improvedmodel, and then make the further improvement on the shortage. At last,this paper putforward a new network model. The new model using cosine function improved thedynamic characteristic of gain function, and improved the noise of the network so asto make it more in line with the change of the Logistic mapping characteristics. Thelast chapter advantage of the new model to solve TSP problem. The simulationexperiments show that the improved model is convergence better than the previousmodel,and get the expected effect. The new chotic neural network model shows itssuperiority.The study is maily based on the following several aspects contents:1. This paper introduces development of chaos neural network,present situation of theresearch systematically, several previous chaos neural network model and it’spotential applications.2. This paper simple introduces rudimentary knowledge of Hopfield neural network,especially Hopfield neural network model, the energy function and its application. It also points out the Shortage that Hopfield neural network is easy to entrap itself intothe local minimum points.3. This paper introduces the definition, characteristic and the chaos to the way of thechaos overall systematically, then emphatically introduce the Logistic mapping theperiod-doubling bifurcation images. It focus on the control parameters value rangeand its change characteristics in the period orbits of the system.4. This paper analysis individual neurons model and several common transient chaosneural network model, analytical model of the deficiencies, improves the algorithmproposed new transient chaos neural network model, and use of in the new model inthe function optimization.5. Will the new transient chaos neural network model is applied to solve the TSP,comparison of its predecessor is introduced and the advanced nature of the networkmodel.