Research On Distribution Optimization And Pricing Decision Of Crowdsourcing Logistics Platform

Crowdsourcing logistics platform is a type of professional internet bilateral platform.The platform connects the employer and receiver to,providing venues and conditions for users of both sides to complete the transaction process of online order and offline delivery,and it has the advantages of reducing operation and management costs,improving distribution efficiency,and providing employment opportunities.In recent years,the crowdsourcing logistics platform has developed rapidly,and at the same time encountered many problems in the development process,mainly including the optimization problem of the platform itself and the problem of competition decision between platforms.In terms of platform itself optimization,distribution network optimization is one of the important factors that affect the development of logistics enterprises.Reasonable and efficient distribution solutions can reduce various costs.The crowdsourcing distribution model has its own characteristics of uncertainty of dynamic orders,multiple starts,open-loop and multiple visits,which brings crowdsourcing logistics optimization of the platform's distribution network the difficulty.Order quantity prediction and distribution route optimization can effectively improve the platform's own service level.In terms of competition decisions between platforms,with the continuous emergence of competitive platforms in the market,the competition between the platforms is becoming more intense.The platform needs to make corresponding decisions to meet the challenges brought by the competition,and platform pricing strategies can effectively help the platform to improve decision-making level in the competition market.Therefore,this paper studies the order forecasting,distribution network optimization and pricing strategies of crowdsourcing logistics platforms respectively.First,study the order quantity forecasting of the crowdsourcing logistics platform.First,collecting time-series data of order quantity and conducting data exploration after data preprocessing.Then,using multiple differences for order quantity time series data,combined with DF to judge the stability of the data after multiple difference,and the best order is selected as the value of the timesteps parameter in LSTM model.Finally,LSTM model is established for training and prediction,and verified by numerical simulation.The study found that the Difference Test makes it more stable and the features of the data can be extracted effectively.Combining it with long-and short-term memory network model training reduces training and test errors,improves the generalization ability of the model,and improves prediction accuracy.Then,study the optimization of crowdsourcing and delivery route considering multiple visits.First,according to the characteristics of crowded logistics distribution,such as open loops and multiple visits,a mixed-integer programming model for crowdsourcing and delivery with the shortest total distribution distance as objective function is established.Then,three-stage heuristic algorithm(H-3)is designed for programming model solving and analyzing.Finally,numerical simulation verifies the effectiveness of the algorithm.The study found that the algorithm in this paper is feasible and effective,and an effective solution that meets the crowdsourcing distribution scenario is obtained.At the same time,H-3 has better performance than traditional algorithms in terms of efficiency improvement,resource-saving and customer satisfaction improvement.Finally,study the pricing of crowdsourcing logistics platforms considering different information levels.First,the information level and direct network externality parameters are introduced to establish the platform profit maximization model,and the platform pricing strategies in monopoly and competitive markets are studied respectively.Then,analyzing the impact of information level and intra-group network externality on the platform's pricing strategy,user size and profit.Finally,the validity of relevant properties is verified by numerical simulation.The study found that,in the analysis of information level parameters,platforms in monopolistic markets prefer higher levels of information,making it easier to set high prices to increase platform profits.Platforms in the competitive bottleneck market prefer lower information levels to mitigate the impact of low-price competition between platforms.In the analysis of direct network externalities parameters,direct network externalities aggravate internal competition,and the size of users at both ends decreases as it increases.In monopoly situations,the platform prefers lower direct network externalities,and in competitive situations,platforms prefer higher direct network externalities.