Logistics Data Predict System Based On Selfadaptive BP Neural Network

The main purpose of the logistics data predicting system in this paper is to study the forecasting problem of logistics order volume and optimize the accuracy of predicting.Through this system,logistics enterprises solve the problem of vehicle scheduling and dispatch in advance and reduce the cost of vehicle transportation.In the process of project cooperation with logistics companies,it was found that the expansion of logistics scale has accumulated massive logistics data after field research.The problem is how to effectively use logistics big data to create more benefits for logistics companies that companies are very concerned about.Therefore,a more efficient and stable forecasting algorithm is necessary to forecast the volume of logistics orders.Firstly,among the data prediction algorithms,the self-applicable BP neural network algorithm has good non-linear data prediction performance and versatility.The parameters are simple and easy to implement,but the algorithm still has some shortcomings.The BP neural network algorithm needs to accelerate the convergence speed of the BP neural network in the training process,and avoid the self-applicable BP neural network from falling into the local optimal value when optimizing the weights and thresholds.The improved particle swarm algorithm has a fast convergence speed so as to avoid falling into the local optimal value.Therefore,this paper combines the improved particle swarm algorithm and proposes a Hybrid Adaptive Back Propagation Neural Network(HABP,Hybrid Adaptive Back Propagation Neural Network),which adjusts the weight and threshold of the selfadaptive BP neural network through an improved particle swarm algorithm.Substitute the optimized neural network weights and thresholds into the self-adaptive BP network for testing,output test data,and adjust the weights of the BP network.It is found that internal turbulence can be avoided and the convergence speed can be improved.Finally,simulation experiments verify the effectiveness of the algorithm model for forecasting logistics data.Secondly,the problem of multi-label classification in logistics data prediction is studied.A logistics data prediction optimization model based on multi-label classification is proposed.Through the use of the HABP algorithm proposed in the previous chapter to obtain the order volume prediction results,combined with the multi-label classification strategy to subdivide and predict the order data.It can improve the subdivision forecast of logistics order data and improve the granularity of predicting,so that enterprises can better understand logistics order data and provide a basis for business decision-making.And through simulation experiments,to verify the effectiveness of the model improvement to improve the prediction performance.Finally,this paper designs and implements a set of logistics data predicting system to provide effective data prediction results and reasonable vehicle scheduling plans based on the optimization of algorithms and models.The system has major functions such as information management,data predicting,and scheduling.And passed the basic system test,verifying that the system meets the design requirements.It further verifies that the algorithm and model proposed in this paper have strong practical value.