Research On Traffic Flow Prediction And Target Allocation Method Based On Stack Self-Encoder

With the continuous growth of economic strength and technology in various countries,especially the rapid development of hard and soft power in recent years,the country's scientific and technological strength and people's living standards have been greatly improved.Despite the continuous progress and development of society,the resources available to people are limited.In people's lives,there are more and more vehicles but road resources are limited.In the national combat readiness,the number and types of enemy targets are increasing,while our firepower platforms and equipment are limited.How to allocate resources rationally has become the focus of all world.Through reasonable prediction of traffic flow and reasonable allocation of target,a reasonable and scientific decision support plan can be designed,which has important theoretical and practical significance for assistant commander making decision.In this paper,stack auto encoder is used to solve traffic flow forecasting and target allocation problems.The main problems and innovations are as follows:(1)In view of the problem that the stack auto encoder network is difficult to converge and easily fall into local optimum due to the high randomness of traffic flow data in traffic flow prediction,this paper designs a neural network algorithm based on balanced exponential smoothing-stack auto encoder,which uses balanced exponential smoothing method to smoothly denoise of traffic flow data.The algorithm is used to predict traffic flow for 5 minutes,15 minutes,30 minutes and 45 minutes respectively,and relevant comparative experiments are designed.Experiments show that the balanced exponential smoothing method in the algorithm can smoothly denoise of traffic flow data of 5 minutes and 15 minutes,accelerate the convergence of stack auto encoder.enhance the generalization ability of network,and solve the problem that the neural network is easy to fall into local optimum..(2)Aiming at the problem that it is difficult to improve the prediction effect of traffic flow with long time interval based on balanced exponential smoothing-stack auto encoder neural network algorithm,it can be seen from the analysis that it is not significant to deal with traffic flow with long time interval from the perspective of data.In view of this problem,this paper analyses and designs stack auto encoder-wavelet neural network algorithm from the aspect of prediction model.The method predicts the traffic flow with four time intervals.The wavelet neural network in the algorithm can express the local information of traffic flow by expansion factor and scaling factor.On the one hand,it can solve the problem of high randomness of traffic flow,which makes the network difficult to converge,and on the other hand,it can effectively extract local information.of traffic flow.The performance of this method in each time interval is relatively high,and a comparative experiment is designed.The experiment proves the effectiveness of this method.(3)Aiming at the problem that it is difficult for fire platforms to quickly formulate reasonable allocation schemes for targets with different attributes and uses,this paper designs a target allocation algorithm based on stack self-encoder-multi-layer perceptron.In the algorithm,a shooting advantage function based on the factors of shooting efficiency and the shortest shooting time by the method of entropy weight is designed,which makes the obtained shooting advantage function more objective.,and the function provides training data for the neural network.The stack auto encoder is used to fuse multi-attributes of aerial targets,which improves the prediction accuracy of deep neural network.The algorithm has the advantages of short allocation time and high strike efficiency,and can be applied to practical engineering projects.