Research On Standardized Container Selection Problem Combining DEA And Bayesian Networks

Logistics transportation is one of the important driving factors of economic and social development,and container freight,as the most advanced logistics transportation method in the world,undertakes 90%of the world’s freight trade.Numerous and complicated containers are not conducive to efficient container freight.Therefore,the rational selection and effective use of standardized modular containers is an important decision-making issue faced by relevant decision-making institution such as the government to improve the operational efficiency of the entire container freight system.Under the development concept of green and sustainable logistics,decision makers of relevant organizations such as the government also need to incorporate sustainabilityrelated factors together with economic and social factors into the decision-making mechanism system for the selection of standardized modular containers.However,the existing related research mainly solves the above-mentioned selection problem of standardized modular containers from the perspective of economic efficiency,and lacks to meet the demands of green logistics under the current social development background from the perspective of green and sustainable factors.On the other hand,some customer-oriented logistics research points out that decisions in real-world operations and management should fully consider customer preferences and behaviors,and a large number of studies show that considering customers’ decision-making behaviors can effectively improve the operational efficiency of logistics systems.However,in the existing research,the dynamic choice preferences of shippers are difficult to effectively describe in the model.And in reality,because the shipper’s decision-making on container selection is very susceptible to external disturbances,it shows a considerable degree of randomness.However,in the research on container selection,the "customer"under the current decision-making system,that is,the shipper’s choice preference,is hardly considered.Therefore,this paper integrates multiple factors such as economy,society and environment to construct a standardized modular container selection model that considers the shipper’s choice preference,so as to effectively solve the current pain points,and finally provide an efficient and accurate intelligent decision-making scheme for the government and relevant decision-making organizations.The main research contents of this paper are as follows.First of all,on the basis of the usual cost-efficiency orientation,this paper uses the DEA method to comprehensively consider the important environmental sustainability factors in the selection of standardized containers and the social factors such as the shipper’s choice preference,and realizes the complex dynamic in the real world.The shipper selection behavior of is accurately characterized in the DEA model of standardized modular container selection.The second part studies the identification and capture of the shipper’s random preference in the more common imperfect rational situation in reality,so as to solve the problem of model failure caused by the uncertainty perturbation caused by the uncertainty preference to the SPM in the first part of the model.In this part,the use of Bayesian network,a machine learning algorithm with high accuracy and high interpretability at the same time,realizes the identification of the random shipper selection preference of shippers under the condition of many intricate and ambiguous factors..And design an efficient Bayesian network result learning algorithm that can deal with the ubiquitous noise and large-scale data efficiently-the dynamic multi-objective discrete artificial bee colony algorithm based on maximum principal subgraph(MPD-DMDABC),so as to achieve Using the shipper’s historical selection data to mine the shipper’s selection pattern of containers to obtain a random preference distribution,and then obtain the shipper’s most likely preference matrix.Finally,the experiment verifies that the shipper’s random preference Bayesian network can accurately and effectively obtain the shipper’s random selection preference.The innovation of this study is as follows.First,this study uses the DEA method to comprehensively consider the economic,social and environmental factors in the selection of standardized modular containers,making up for the sustainability factors such as energy consumption and environmental pollution that are ignored in the existing research.At the same time,it effectively solves the problem of bias caused by weighted or unweighted treatment of multiple factors as exogenous variables in traditional research.Second,this study successfully solves the problem of effectively characterizing the complex and dynamic real-world shipper choice behavior in a standardized modular container selection model.This study is likely to be the first time so far to incorporate the factors of shipper choice preference in standardized modular containers.Select studies that characterize effectively.Third,this study mines the shipper’s preference pattern from the shipper’s selection data to obtain the result of random selection of shippers under imperfect rationality.It effectively solves the problem of disturbance and uncertainty of the SPM matrix caused by the random preference of non-fully rational shippers in reality.This indeterminate SPM would render the above standardized modular container selection model unusable.Fourth,the Bayesian network learning algorithm designed in this study,MPD-DMDABC,can more accurately and efficiently learn from the current ubiquitous noisy and large-scale data compared with other existing algorithms,and can dynamically adapt to different demand conditions.