Research On Differential Evolutionary Integrated Control Algorithm For Rail-Guided Vehicle

With the booming development of industrial Internet and Internet of Things related technologies,the collaborative manufacturing of rail-guided vehicle(RGV)based logistics management devices and multiple computer number controls(CNC)has become a hot spot and a difficult research area.Through the integrated control of CNCs,efficiency improvement can be achieved in the production workshop scenarios of key parts in manufacturing industries such as aviation,shipbuilding,and automotive.This research problem can be called RDSP(RGV dynamic scheduling problem).At present,for the RDSP,the difficulties mainly lie in: first,collaborative control is difficult to achieve.Unlike the conventional shop scheduling problems,in the RDSP,the control algorithm must consider not only the sequential process of materials by CNCs in each process,but also the flow process of materials in the job shop.Second,the search efficiency is difficult to improve.Considering the efficiency of the production process,the continuous processing time of CNC is usually long,which also leads to an extremely large solution space and poor search efficiency.Third,the load difference is difficult to reduce.In the machining process,the CNC with the largest load will have a decisive impact on the average completion time of the materials,which ultimately makes it difficult to control the machining loss.The contributions of this thesis mainly include:(1)In order to solve the problem of difficult collaborative control,a collaborative optimization mechanism for CNC assignment and RGV scheduling is designed.A new coding mechanism is designed to achieve independent coding and collaborative optimization for CNC process assignment and RGV scheduling control.In this case,the process assignment corresponding to the CNC code determines the upper limit of the finished material output of the workshop,while the scheduling strategy corresponding to the RGV code determines the productivity of the machining system under the above limit.(2)In order to solve the problem of low search efficiency,a multiple population coevolution mechanism based on transdifferentiation strategy is designed.In each iteration,the generated offspring population will be divided into superior,impaired and eliminated populations according to the fitness level of the individuals.The superior population is responsible for maintaining the superiority of individuals,while the impaired population is responsible for maintaining the diversity of the whole population.In the impaired population,individuals are "harmed" according to their fitness and some decision variables of them are regenerated.(3)In order to solve the problem of large load variation,an RGV code generation mechanism based on the debilitating factor is designed.The corresponding RGV codes can be generated adaptively based on the processing demand of the material and the mechanical characteristics of the equipment.In this process,based on the decreasing distribution determined by the debilitating factor,the number of state transfers of the Markov process in the material flow process is derived,and optimization is carried out under this limit,thus reducing the optimization difficulty of load balancing.To solve the above problems,a theory-driven integrated differential evolutionary control algorithm for RGV,called DE-TS,is proposed in this thesis.In this thesis,the superiority of the proposed algorithm DE-TS is verified in various aspects through simulation experiments.First,the effectiveness of the introduced optimization strategy is verified by eliminating the introduced optimization strategy and finding that the system loss increases by 45.8% and 46.9% for different test problems and CNC number scenarios,respectively.Secondly,by comparing with other algorithms on nine test cases,it was found that the machining efficiency achieved by DE-TS was 25.68% higher on average,thus verifying the superiority of the DE-TS in terms of performance.Finally,the excellent performance distribution possessed by the DE-TS is verified by sampling the algorithm performance on the extended set of standard experiments.In order to analyze the intrinsic reasons for the superiority of the proposed algorithm and its stability level,convergence analysis of the DE-TS algorithm and its interpretability was carried out.By analyzing the distribution of the optimal solutions obtained from the repeated experiments,it was found that the lengths of the intervals from the lower quartile to the upper quartile of the obtained loss were 3.01%,0.34%,and 0.25%,respectively,thus verifying that the DE-TS has a high degree of convergence.Subsequently,by performing an interpretable analysis of the transfer probability matrix of the Markov process of the algorithm with the RGV material transfer path,it was found that the DE-TS can indirectly balance the processing load levels among different CNCs by controlling the RGV material transfer path,revealing the intrinsic reason for the superiority of the proposed algorithm.This research is expected to provide technical guidance for the machining production of critical parts in important manufacturing fields such as shipbuilding and aerospace.