Research On Intelligent Assembly Sequence Planning For Complex Assembly

Assembly sequence planning(ASP)is an important part of virtual manufacturing technology.Its results affect the arrangement of assembly process and the accuracy,quality and performance of final products.Based on the problems existing in ASP of complex products at the present stage,the method and key technology of ASP have been researched.A method of assembly sequence planning and optimizing for complex products based on the hybrid cycle algorithm and virtual reality technology was proposed in this thesis.In the stage of assembly sequence planning,a new algorithm,hybrid cycle algorithm,was proposed.This algorithm is based on the genetic algorithm,combined with the hierarchical model of product structure and the same coding method of the same kind of parts to improve the problem of complex arrangement and combination caused by too many parts.The interference matrix and contact matrix are used to adjust the randomly generated assembly sequence to avoid the problem of reduced operation speed caused by the calculation of invalid arrangement and combination fitness.In order to overcome the problem that genetic algorithm is easy to fall into the local optimal solution,the annealing idea of simulated annealing algorithm and the Metropolis criterion were introduced to accept the deteriorating solution probably.The idea of tracking the extreme value of particle swarm optimization is introduced to record the optimal sequence and directly cross and mutate with the offspring so as to improve the speed of sequence optimization.Finally,the fitness function is constructed by the two indexes of the unity of assembly direction and the unity of assembly tools to realize assembly sequence planning,and taking automobile rear axle as an example to verify the effectiveness of the hybrid cycle algorithm.In the stage of assembly sequence optimization,we proposed a sequence optimization method to simulate product assembly in virtual environment.Firstly,a new collision detection algorithm named stereo model stacking algorithm was proposed to achieve the accessibility of assembly parts and the maneuverability of assembly tool space.The algorithm was based on the generating principle of stereo model in unity3 d software.Firstly,the triangle sequence array and the vertex coordinates of each triangle patch were extracted from the grid components,and then each triangle patch equidistant is divided from the vertical and horizontal dimensions according to the set precision value.Cubes with side lengths as accuracy values were generated by using each segmented node as the center to form a part shape.Built-in box colliders were added to all cubes to implement a stacked model instead of the part body for collision detection,which could finally solve the problem of insufficient collision detection accuracy for irregular parts in virtual assembly.Finally,the virtual reality assembly simulation platform of automobile rear axle was established to fully utilize the advantages of manual assembly experience,so as to solve the problem of poor sequence optimization caused by the lack of deep human-computer interaction in the past.The platform took the automobile rear axle as an application example,including six main functions: sequence optimization,assembly demonstration,working principle,free disassembly,multi angle section and parts details.It was equipped with HTC VIVE Pro virtual reality equipment runs,and the handle controller was used to realize the purpose of disassembling,assembling and analyzing the structure principle of the automobile rear axle products according to the assembly sequence obtained by the algorithm.All of the above,we could further optimize the assembly sequence in a more intuitive and realistic way.