Research On Logistics Distribution Routing Optimization Algorithm Based On Deep Belief Network

Research on Path optimization problem has been a long history, early problems of complexity was low, which can be solved by mathematical modeling of the exact algorithm. Subsequently, the computational complexity of the algorithm becomes higher and its cost time gets longer as the scale of problem increasing. The evolutionary algorithm was merged into the path optimization, in which it is called heuristic algorithm, being a kind of empirical algorithms. Although it cannot guarantee the optimal solution, an acceptable solution can be obtained during a relatively short term. Path optimization plays a significant role in the logistics distribution. Albeit many works existing, a number of researches has still been in the stage of simulation theory, and heuristic algorithms take no effect on the evaluation of the actual logistics distribution path.With the vigorous development of China's logistics industry, the logistics industry is in the urgent need of relevant theoretical and technical support. In recent years, China' s urban traffic congestion problem is very serious, so logistics distribution path optimization is not a simple combination of optimization problems. Because the complex changes of road traffic have great influence on the optimization of logistics distribution route, the route passage time has great ups and downs in different time periods, and the research on the optimization problem of logistics distribution pathways should consider more problems of road conditions.In order to solve the problem that the traditional heuristic algorithm is not suitable in the practical logistics distribution and the complex road conditions incurs poor-qualitied results, this paper presents a DBNTF (Deep belief network traffic forecast) road prediction algorithm based on the deep belief network model.Firstly, the traffic forecast model DBNTF is constructed. which employs five layers of DBN (Deep Belief Network) structure and softmax classifier. Then the traffic data is preprocessed, the pre-processed traffic data set is used as the input layer of the DBNTF model, and the traffic level label is set manually to complete the training of the model. Finally, the predicted traffic conditions (traffic level value) can be obtained through entering the weather data and other characteristics of the data. On this basis, the traffic grade value is transformed into time-sharing section weights, the time-sharing section weights and traffic network constructing the time-sharing traffic network .According to the characteristics of urban logistics distribution and the demand of algorithm, the ant algorithm is modified and combined with Time - sharing traffic network, then DBNTFPO (Deep belief network traffic forecast path optimization) algorithm is proposed. In the experiment of comparing the two algorithms and other algorithms, the results show that the superiority and value of application. In the actual logistics and distribution,we provide a feasible and effective idea and approach for path optimization issue,which exerts a positive impact on logistics and traffic guidance.The main contribution of this paper is as follows:1) In this paper, the traditional evolutionary algorithm in the logistics distribution path optimization in the principle of research, analysis, and introduces the deep learning technology for the shortcomings.2) The traffic forecasting model is constructed and the traffic prediction algorithm-DBNTF algorithm is proposed based on the traffic forecast model. The accuracy of the algorithm is verified by the algorithm in complex environment.3) The logistics distribution path optimization model is constructed, and the algorithm of logistics distribution path optimization - DBNTFPO is proposed according to the model, and the superiority of the proposed method in predicting the accuracy of complex road condition is verified.