Research On Multi-specification Cargo Loading Based On Improved Random Forest Algorithm

The rapid development of economy stimulates people's demands for life.In add ition,with the continuous popularization of Internet and online shopping,the types and quantity of goods purchased online are constantly increasing,which means that under the general trend of Internet,a large number of goods of different specificat ions will be transported.The increasing types of cargo transport brings great pressu re to container transport.Three-dimensional packing problem is a typical NP-ha rd problem.Many researchers have done a lot of research on this problem,but it i s still difficult to get an effective solution.In recent years,some researchers have proposed some algorithms to solve the 3D packing problem,and achieved good res ults.The improved random forest algorithm is to replace the CART tree part in the advancing forest with the optimization rule of the new extremely fast decision tree algorithm,so the computation time of the algorithm is greatly shortened.In this pa per,the improved random forest algorithm is applied to the 3D packing problem.T he algorithm firstly Bagging the data of goods to be loaded into the set decision tr ee number.Secondly,for each group of data of goods to be loaded,the optimal ch aracteristics are determined first,and then each loading layer of the container is reg arded as a tree.Finally,the loading with the maximum eigenvalue is selected from the goods to be loaded in line with the remaining space of the container.The rem aining space on the upper and front of the newly loaded box should be updated fo r each loading to avoid space waste and ensure the space utilization rate of the co ntainer.Because of the low complexity of the improved random forest algorithm an d the high speed decision tree to improve the algorithm time,and because of the h euristic search algorithm to ensure the utilization of container space,can get a bette r packing effect,so the study of this paper has a high research value.The main research contents of this paper are as follows:first,through the resea rch of three-dimensional packing problem,the packing constraint conditions are dete rmined,and the corresponding mathematical model is established.Secondly,an impr oved stochastic forest algorithm is used to solve the 3D packing problem.Firstly,t he basic idea and algorithm flow of integrated learning model,decision tree,fast d ecision tree,heuristic algorithm and random forest algorithm are introduced.The ap plication and implementation steps of improved random forest algorithm in container loading are designed.Finally,the algorithm is implemented by programming.Third,fused by heuristic search algorithm and fast decision tree to improve the random f orest algorithm using BR1-BR10 this 1000 instances experiment test,and to GRA SP,the Maximal-space,HSA,CLTRS,MLHS,FDA,earth-sized also use these in stances algorithm to get the results using the comparative analysis,verified the feas ibility and validity of this algorithm.The conclusion of this paper is as follows:in this paper,1000 examples of BR 1-BR10 are used to test the improved random forest algorithm,and the test results are compared with the classical packing algorithm at present,which proves the effe ctiveness and feasibility of using the improved stochastic forest algorithm to realize the 3D packing problem.However,this algorithm still has some shortcomings,suc h as low container utilization.Figure 16 Table 13 Reference 91...