Research On Scheduling Optimization Algorithm Of Intelligent Factory Based On Capacity Balance

With the development of industrial Internet,the network in the factory presents three major development trends: convergence,openness and flexibility.The workshop-level network and field-level network in the factory are gradually integrated,and high real-time control information and non-real-time data can be effectively transmitted through a common network.The new network technology represented by time-sensitive network breaks many technical barriers of traditional industrial network and promotes the real-time transmission of industrial process data.Thanks to a variety of technologies,the intelligent factory control platform is able to bridge the transmission of information with on-site industrial equipment.The existing manual scheduling optimization method can no longer match the speed of information acquisition,so it is necessary to carry out intelligent scheduling.However,the factory production situation is complex and the production rhythm is fast,so it is still a research hotspot to develop a scheduling optimization algorithm which is close to the reality of the factory and can converge quickly.At the same time,with the rapid development of intelligent factories,capacity bottleneck has gradually become an important factor restricting the development of factories.How to make effective use of on-site data and information while scheduling optimization is also an urgent problem to be solved.In order to solve the above problems,this thesis proposes an optimization algorithm for intelligent factory scheduling based on capacity balance through the research of traditional workshop scheduling and multi-objective problems.The specific research contents are as follows:(1)In this thesis,a smart factory scheduling optimization model is proposed for the actual production situation with complex and changeable order structure,unknown input order type,and high convergence time requirements.Firstly,based on the traditional job shop scheduling model,a basic model suitable for the actual situation of intelligent factory is established,and the concepts of order job and production line slot are introduced to cut and distribute orders dynamically,by changing the minimum allocation unit of products.As a result,the time complexity of solving the scheduling optimization model is reduced.Then,for the improved model,we designed the corresponding external penalty function to eliminate infeasible solutions,and designed the solution process of the model,using a variety of neighborhood search algorithms to solve the model.The experimental results show that the model solution results are effective,and the model is stable in the solution process of various algorithms,while the solution time of the traditional workshop scheduling model and its improved model is longer than this model.(2)On the basis of the improved smart factory production scheduling optimization model,this thesis proposes a smart factory production scheduling optimization algorithm based on capacity balance for the situation that smart factory decision-makers are inexperienced,the order composition structure is complex,and the production lines are relatively independent.Firstly,three production capacity indicators are established according to the factory situation,and these three indicators are used as objective functions,and a dynamic normalization method is designed for each objective function.Then,based on grey comprehensive correlation analysis and single objective optimization,a method of using pre-scheduling analysis to analyze the relationship between various capacity indicators is designed,so as to provide the weight of suggestions for decision-makers and balance each capacity index,realize the optimization of intelligent factory scheduling based on capacity balance.Finally,the effectiveness of the algorithm is verified,and the results of the algorithm are compared with the single-objective optimization results and the multi-objective optimization results based on the entropy weight method to determine the weight.The experimental results show that at the overall solution set level,the algorithm is closer to the single-objective optimal solution set than the entropy weight method in most data sets;at the individual optimal solution level,the algorithm has a stronger ability to balance the productivity index of largescale data than the entropy method.