Research And Application Of Deep Synergetic Neural Network Based On Synergetics

Deep learning is one of the most popular topics in science and technology,and it is also the most successful and effective method of thinking in the field of artificial intelligence.Research and application based on deep learning theory are also endless.However,with the deep study of deep learning,deep learning areas exposed by the high cost of computing and training costs are too high and other issues need to be resolved.This paper mainly studies the deep synergetic neural network based on synergetics principle,which is a new deep learning network model,which can effectively reduce the high cost of calculation and the high training cost in traditional deep learning area.The main contents of this paper are:First of all,it introduces the basic idea,mathematical model and related important concepts of synergetic theory.Then we introduce a new class of neural network model based on synergetic theory: synergetic neural network.This paper introduces the mathematical model,structural model and operation flow of synergetic neural network,and introduces several basic algorithms of synergetic neural network: classifier algorithm based on PFR model,classifier algorithm based on PFAP model,SCAP algorithm and SCAPAP algorithm.Detailed description of synergetic and synergetic neural network of the various characteristics.Secondly,it introduces several traditional deep neural network models: convolution neural network and deep belief neural network,and introduces their model structure and operation process respectively.Based on the synergetic principle and the model of structure traditional neural network,the deep synergetic neural network is constructed,and the model structure,the running process and the algorithm steps of the deep synergetic neural network are described in detail.Which provides sufficient theoretical support for the research and application of the subsequent deep synergetic neural network.Finally,based on the above-mentioned deep synergetic neural network model,the performance experiments of deep synergetic neural network under different sample banks are designed.At the same time,the deep synergetic neural network is compared with the traditional convolution neural network and deep belief neural network in the same sample library Under the experimental test,the longitudinal comparison of their performance characteristics.The results of comprehensive experiment show that the deep synergetic neural network not only has good performance in recognition effect,but also has better performance in calculating cost and running efficiency.Based on the performance characteristics of the above-mentioned deep synergetic neural network,this paper also designs a flight control system that uses the deep synergetic neural network to identify the declarable landmarks during the autonomous landing of UAVs.Based on the simulation data and the identification of autonomous landing signs,the validity of the proposed deep synergetic neural network model is verified,which lays the foundation for further research on UAV autonomous landing and air obstacle avoidance.