Imitated Ecology Neural Network Algorithm Research And The Application In Sound Signal Vehicle Type Recognition

The neural network is based on the physiology biological modelling model, more and more applications are used in each domain, especially the pattern recognition in recent years. But artificial neural networks' fundamental research mainly focuses on the human brain biology neural network information storage and the processing mechanism simulation, people pay little attention to its community operation way. Learning from the biology angle, the human brain biology neural network which is formed by the great quantity nerve cell's widespread interconnection is carrying on the community operation by nervous system's form. Thus it is vital significant to study the relations of the multi-neural network.Compared the artificial neural networks model with the cerebrum, we may discovery that obviously, the function which all artificial neural networks model has is only is minimum part of the cerebrum.The brain structural research indicated that humanity's intelligence is accumulates depending on the education and the experience forms. The human brain is divided into different zones, for example the feeling area, remembers the area, the language area, literate area, the motor area and so on. Different zones execute different function, some manage vision, some manage language, and some manage pondered. But when cerebrum studies new knowledge, it is a knowledge accumulation process, does not need to learn the knowledge again, which may realize the knowledge expansion. Therefore, by joining many known function artificial neural networks together, composing a consummation function large-scale artificial neural networks system is completely possible.Proposed the imitated ecology neural network new idea based on the neural network integration's thought and the characteristic and ecosystem's model in this article. I mainly research and compilation imitated ecology neural network fusion learning algorithm and training regulations production process. I also have compiled the imitated ecology neural network procedure with Matlab6.5, and use the imitated ecology neural network in the automotive type classified recognition, compared with the single neural network recognition; it raises the vehicle classification accuracy, and cut in the approximate vehicle recognition probability of misrecognition.