Research On Mechanism Kinematic Chain Isomorphism Recognition Based On Hybrid Discrete Particle Swarm Algorithm

Mechanism kinematic chain isomorphism identification has always been a key issue in the synthesis process of kinematic chain structure.In the mechanism design,the isomorphic motion chain should be removed to improve the design efficiency.Since this problem has proved to be a nondeterministic polynomial(NP)problem and has not been effectively solved,finding an efficient and accurate method for mechanism kinematic chain isomorphism identification has always been one of the goals of researchers.Due to its good robustness and parallelism,intelligent algorithms are widely used in the field of practical engineering and have become one of the hot spots in the interdisciplinary research field.Therefore,this paper simplifies the form of the problem by transforming the complex mechanism kinematic chain isomorphism identification problem into an optimization problem of the algorithm through mathematical description.The main work is as follows:(1)The theory of graph theory is applied to mechanism science,and the topological diagram is used to represent the relationship between the components and motion pairs of complex motion chains.According to the isomorphism definition of the graph and the mathematical description of the motion chain,the adjacency matrix of 0,1 is used to represent the association between the edges and vertices of the topological graph,and a mathematical model for the determination of mechanism kinematic chain isomorphism identification is established,which simplifies the expression form of the motion chain and lays a foundation for the design of the subsequent fitness function.(2)An improved discrete particle swarm algorithm is proposed and used to solve the problem of isomorphism recognition of mechanical kinematic chains.Firstly,based on the characteristics of multi-dimensional adjacency matrix and isomorphism recognition criterion,the method of particle position and velocity update are redefined.In order to solve the problem of initial population randomization and unstable operation of the algorithm,the reverse learning mechanism is introduced to improve it to achieve the purpose of optimizing the initial state of the population.Secondly,in order to strengthen the simulation optimization performance of the algorithm,the adaptive cosine change formula is introduced to optimize the parameters of inertia weight and learning factor,which enhances the global search ability of the algorithm.Then,aiming at the easy early convergence in the late iteration of discrete particle swarm algorithm,Cauchy mutation strategy is used to carry out particle disturbance,and the extension of both ends of Cauchy distribution can expand the search range,which is conducive to the development of population diversity,and its mutation mode effectively improves the situation that discrete particle swarms are easy to fall into local optimum.Compared with the Hopfield neural network,taboo search,and unadjusted discrete particle swarm algorithm,the improved discrete particle swarm algorithm shortens the average operation time by about 50%.The results verify that the proposed improved discrete particle swarm algorithm has certain feasibility and efficiency in solving mechanism kinematic chain isomorphism identification.(3)When dealing with cumbersome members,the original defects of a single particle swarm are not suitable for the mining and exploration of big data,and a hybrid discrete particle swarm-artificial fish swarm algorithm is proposed based on this to try to change the shortcomings of the single algorithm.The two algorithms complement each other to a certain extent,thereby effectively improving the performance of the entire system.The group optimal value of discrete particle swarm is used to optimize the initial state of artificial fish group,and the individual optimal value is introduced into the clustering behavior of artificial fish group,which improve the search efficiency of the hybrid algorithm.At the same time,dynamic inertial weights are used to optimize the field of view of fish schools,so that the algorithm has directionality and target,which improves the convergence rate of fish stocks.The results show that the improved hybrid algorithm is significantly shorter than other algorithms in terms of operation time,and the correct rate of judging whether the motion chain is isomorphic is almost100%.The simulation results verify that the proposed hybrid algorithm has better optimization performance.