Mathematical Modeling And Intelligent Algorithm For Man-robot Cooperative Two-sided Assembly Line Balancing Problems

Assembly line production lines are widely utilized in modern manufacturing industry to produce the standardized products.As a new assembly line layout,two-sided assembly lines are utilized in automobile,truck,motorcycle and other manufacturing industries due to the advantages of shorter assembly line length,lower equipment cost,lower material handling cost,etc.With the increase of human worker cost and the support of the Chinese government,robots are being gradually applied to two-sided assembly lines to partially or totally replace manual labor,resulting in robotic assembly lines or man-robot cooperative assembly lines.In robotic two-sided assembly lines,the cost of energy consumption is an important part of the robot's daily running cost.And the optimization of energy consumption cannot only reduce the cost of enterprises and improve the competitiveness of enterprises,but also reduce environmental pollution and carbon emissions effectively.As there are different assembly situations,this research gradually studies the standard two-sided assembly line balancing(humans operate the tasks),robotic two-sided assembly line balancing,and energy-efficient robotic two-sided assembly line balancing,and finally studies the man-robot cooperative two-sided assembly line balancing by solving realistic cases.In view of the standard two-sided assembly line balancing problem,this research develops new more efficient decoding schemes and develops an improved local search algorithm.This research firstly studies the encoding schemes and decoding procedures,and develops more effective decoding methods to balance the workloads on the two sides of a mated-station and reduce the sequence-dependence idle time.In addition,this research develops new heuristic objective to preserve the task allocation with less idle time on the former mated-stations.On the basis of the improved decoding procedure and heuristic objective,a neighborhood structure based iterative greedy algorithm is developed,which proposes effective heuristic initialization and a new local search based on two kinds of neighborhood structures.In order to evaluate the proposed decoding schemes,the proposed two decoding schemes are compared with the other seven decoding schemes,and the computational results show that the proposed decoding methods have the best performance.To test the performance of the proposed algorithm,it is compared with other 14 algorithms,and computational results validate that iterative greedy algorithm has the best performance under four termination criteria.Meanwhile,the proposed algorithm updates the best solutions of the 12 cases,and the number of updated cases by this algorithm is the maximum among that by the tested algorithms.Regarding robotic two-sided assembly line balancing problem,this research constructs a mathematical model and develops swarm intelligence algorithms to solve this problem fast.A mixed integer programming model is developed to minimize the cycle time and solve the small-size instances optimally.As the considered problem involves two sub-problems,task assignment and robot allocation,this research proposes task permutation vector and robot allocation vector for encoding,and develops an effective decoding procedure based on the study on the encoding schemes and decoding procedures of the standard two-sided assembly line.Subsequently,a discrete cuckoo search algorithm and a cooperative coevolutionary cuckoo search algorithm are developed to optimize the task assignment and robot allocation simultaneously.To evaluate the performance of the proposed algorithms,they are compared with other five algorithms.The experimental results show that the discrete cuckoo search algorithm shows better convergence,and the proposed coevolutionary cuckoo search algorithm has the best performance under three termination criteria.As for energy consumption optimization in robotic two-sided assembly line,this research develops a mathematical model and multi-objective optimization algorithms to optimize energy consumption.The multi-objective optimization model is constructed to minimize the cycle time and total energy consumption simultaneously.As the two objectives might be conflicted in some occasions,this research develops new decoding scheme and decoding problem on the basis of robot allocation vector,task assignment vector and task permutation vector,and then proposes fast non-dominated genetic algorithm and multi-objective restarted simulated annealing algorithm to optimize the two objectives simultaneously.Computational results on a set of benchmark problems show that the energy consumption cannot be reduced effectively when only optimizing cycle time,and multi-objective algorithms are capable to achieve many schedule schemes with lower power consumption for production decision makers to select,which further verifies the rationality of the multi-objective model.Meanwhile,the experimental results validate that the restarted simulated annealing algorithm is superior to several original simulated annealing algorithms,and it outperforms non-dominated genetic algorithm in both convergence and spread criteria.The above research findings are directly applied or adjusted to solve the cases taken from the actual automobile assembly lines by combining the above theoretical results and production practice.The actual production status of a car assembly line exhibits that there are three assembly situations: human-oriented assembly,robot-oriented assembly and human-oriented and robot-oriented assembly in the automobile assembly workshop.On the basis of the above findings,this research investigates three cases in order to further study the actual production situations.The computational results demonstrate that the proposed model and algorithms are capable to reduce the cycle time,improve the assembly efficiency and reduce energy consumption effectively,and thus have a good application prospect.