| For different regions,different production scales and production chain needs,agricultural equipment of small batch,multi-species,customized mixed flow production has become an important feature of the current agricultural equipment production,but such mixed flow multi-species production has led to the complexity of production scheduling.At present,agricultural equipment production scheduling is manual to carry out,and can not guarantee the quality of the production plan,which will have a greater impact on the enterprise’s production efficiency.To address the above problems,this paper takes the mixed-flow multi-species production of tractors as an example and establishes the objective model of production scheduling by analyzing the enterprise production management process.After that,the target model is solved by intelligent optimization algorithm and the production scheduling scheme is obtained.In order to ensure the reliability of the production scheduling,a digital twin assembly line is constructed to simulate and optimize the production scheduling scheme.The main research contents and results of this paper are as follows:(1)For the order decomposition problem in production scheduling,an optimization model of order decomposition is constructed based on four factors,including order delay cost,inventory cost,downtime cost,and overtime cost,as the objectives,and the model is solved using the particle swarm optimization(PSO)algorithm.The results show that the optimized decomposition plan is lower in total downtime cost,total overtime cost and total inventory cost,and the total additional cost is reduced by 5998 yuan,which is 15.9% lower than the unoptimized decomposition plan.(2)A multi-objective sequencing model with the objectives of minimizing the total completion time,minimizing the total line stopping time and minimizing the number of product type switching is developed for the mixed-species assembly line of tractors that contains both synchronous and asynchronous assembly lines.An improved virus immune optimization algorithm(ICHIO)is proposed to solve the multi-objective model by introducing an adaptive social distance formula to balance the search ability of the algorithm in the early and late stages,introducing a cross-variance operator to enhance the global and local search ability,and adding a Pareto merit grading mechanism to strengthen the algorithm’s ability to solve the multi-objective problem.The analysis of the results of different algorithms shows that the total completion time of the optimized production sequence is reduced from 22,458 s to 17,778 s,the total stopping time is reduced from 5830 s to 622 s,and the longest stopping time in each stage is reduced from 874 s to 97 s.Meanwhile,the optimized production sequence also takes into account the number of product type switching,which ensures the centralized production of the same type of products to a certain extent.(3)To address the problem of large differences in the working hours of workers on the production line due to the assembly of different models,a picture library of each working process was established by collecting videos of workers’ operations,and a process state detection model was constructed based on Faster-RCNN convolutional neural network.The analysis results of the field video show that the average detection mean(map)of the model detection reaches 96.4%,and the relative error of the working hours recognized by the video is within 10%,which meets the relative error requirement of the actual working hours observation in the production line.(4)For the construction of the digital twin assembly line,3D modeling was carried out according to various types of equipment in the tractor assembly line,and 3D scenes of the workshop paint front line and paint back line were built according to the actual arrangement of each work station in the production line.The digital assembly line human-computer interaction process such as 3D scene information board and role view switching control were designed.The twin assembly line model was driven in real-time according to the real-time over-point information in the production line MES system.(5)Production planning simulation optimization based on the digital twin assembly line is proposed to address the difficulty of verifying traditional scheduling methods through actual production.A production planning management system is developed to integrate order identification,order decomposition,and production sequencing functions.A data transfer protocol between the production planning management system and the digital twin assembly line is established using Rabbit MQ,which allows the production plan to be sent to the digital twin assembly line,and the actual work hours of the production line are used to drive the production plan simulation and evaluation.The production plan was evaluated and analyzed using historical production scheduling data,and the results showed that the sum of production plan completion time,the sum of line downtime,and the sum of overtime or downtime within 5 days for each of the optimized low,normal,and high production months were smaller,and there was a greater reduction in total extra cost. |
