Research And Application Of Improved BP And Genetic Algorithm

The neural network is a interdisciplinary studies and develops very rapidly, which is composed of a large-scale adaptive nonlinear dynamic systems by the massive processing units. The neural network has the distributional memory, the parallel processing, the high fault-tolerant ability and good self-learning, adaptive, association and so on. For this reason, the neural network has attracted many researchers’ interest, its theoretical result and the application achievement emerge one after another. Now it has proposed many kinds of neural network model, one of the most widely used is the BP neural network. The multilayer perception, trained by the back propagation algorithm is currently the most widely used neural network, the essence of back propagation networks is that make the change of weights become little by gradient descent method and finally attain the minimal error.The quantum genetic algorithm is based on the principles of quantum computing and is one kind of genetic algorithm, it is an optimization method which combines quantum computation with genetic algorithm, and it has the characteristic of population with small-scale, fast convergence, globe optimal searching ability. Quantum genetic algorithm essentially is a genetic algorithm, so traditional genetic algorithm can be applied fields, quantum genetic algorithm is also applicable. Because it has introduced the quantum computation, its effect was superior to the traditional evolutionary algorithm.In order to overcome the shortcomings that BP algorithm is usually trapped to a local minimum and a low speed of convergence weights, according to the advantage of the globe optimal searching of Quantum genetic algorithm, this paper proposes a new training neural network mix algorithm-QGA-BP algorithm, and try to put this method used in computer network security study. This algorithm can improve the convergent rate and convergent precision by comparing the GA-BP and BP algorithm and analyzing the results of real examples.The following are the main research contents of this thesis:Firstly, it shows a systematic and detailed introduction to the BP neural network and the algorithm, analysis of its major shortcomings and their causes. In view of the present BP algorithm’s insufficiency, it proposes a new training neural network mixed algorithm-QGA-BP algorithm.Secondly, it presents a quantum genetic algorithm to the BP algorithm. According to the operator of classical genetic algorithm, it add quantum cross operator and quantum mutation operator into the quantum genetic algorithm. Using the improved algorithm to the BP algorithm, the convergence speed of the algorithm is faster, and can prevent the algorithm getting into the local optimization.Thirdly, The BP algorithm, the GA-BP mixed algorithm and the QGA-BP mixed algorithm are separately applied to the experiment of computer network security evaluation, and then compare the advantages and disadvantages of the three algorithms through the experiments.