Research And Implementation Of Flexible Scheduling For Multi-objective Optimization Of Industrial Internet

The Industrial Internet is a critical national development strategy that plays a pivotal role in industrial manufacturing.By enhancing production efficiency,improving product quality,and increasing customer satisfaction through digital means,the Industrial Internet is of tremendous importance in promoting the digital and intelligent transformation of the manufacturing industry.As the global economy continues to evolve,the Flexible Job Shop Scheduling Problem(FJSP)has emerged as a fundamental optimization issue in the manufacturing industry,given the growing demand for personalized and diversified orders.However,current production scheduling methods are predominantly manual,leading to inefficiencies and an inability to meet customer needs.On the one hand,due to the single optimization goal considered in manual production scheduling,carbon emission indicators are not considered;on the other hand,due to frequent dynamic disturbances in the manufacturing workshop,it has a great impact on production scheduling.To address the aforementioned issues,this thesis investigates a multiobjective optimization algorithm and proposes a high-dimensional,multiobjective optimization algorithm,NSGA-Ⅲ-QPBI,to solve the highdimensional multi-objective FJSP.Additionally,a robust scheduling scheme is proposed to improve the stability of the scheduling scheme.Finally,an smart planning and flexible scheduling system is designed and implemented,which automates scheduling in the workshop and resolves the low efficiency and slow response times associated with manual scheduling.The primary contributions and innovations of this thesis are summarized as follows:a)NSGA-Ⅲ has demonstrated notable success in addressing multiobjective optimization problems.However,its performance sharply declines when the number of objectives to be optimized exceeds three.Furthermore,the algorithm exhibits slow convergence speed and inadequate diversity.To address these challenges,we propose an enhanced algorithm,namely NSGA-Ⅲ-QPBI.Firstly,to enhance the convergence of the algorithm,the PBI method from MOEA/D is introduced,and the penalty parameters are adaptively adjusted by reinforcement learning.The convergence and diversity are dynamically balanced according to the current distribution of the target space.Next,to address the issue of poor diversity in high-dimensional space and the decrease of diversity caused by the introduction of PBI in NSGA-Ⅲ,the niche elimination method is introduced.Individuals from different dominance layers participate in the elimination process,with individuals with good diversity in the marginal layer being retained to improve the diversity of the algorithm.Finally,to achieve low-carbon,low-energy workshop scheduling,carbon emission indicators are introduced,and the five indicators of completion time,carbon emission,machine load,delivery time and production cost are simultaneously optimized and solved.The performance of the proposed algorithm is compared with that of other algorithms for solving the FJSP standard sample and the SMT workshop sample,demonstrating better convergence and diversity.b)Machine failures are a common occurrence in real-world workshops,and both pre-reactive and fully reactive rescheduling can have detrimental effects on the original scheduling scheme.Thus,this thesis proposes a robust scheduling scheme that aims to enhance the stability of the initial scheduling scheme during rescheduling.Specifically,two robustness indexes are introduced,which are optimized concurrently with the five existing indexes to generate a more resilient scheduling scheme.Moreover,a machine failure simulation method is proposed to predict when machines fail and when machines are repaired,thus offering a sound basis for robust scheduling.Through experimental comparison,the effectiveness and superiority of this method for solving dynamic FJSP are verified.When solving the SMT workshop sample disturbed by machine failure,a scheduling scheme with better stability and better performance is generated.c)Currently,the SMT workshop still relies on manual production scheduling,which is known to be inefficient,challenging to allocate resources reasonably,unable to respond to customer delivery dates accurately and quickly,and difficult to handle dynamic disturbances.To address these issues,a smart planning and flexible scheduling system has been designed and implemented.The system integrates the highdimensional multi-objective flexible job shop scheduling algorithm and the robust scheduling algorithm proposed in this thesis to improve production scheduling efficiency,accurately predict delivery times,and stably handle dynamic disturbances.Next,the system’s demand analysis,outline design,and development testing were carried out.Finally,the system was deployed online,passed factory acceptance after three months of trial operation,and is currently running stably in the mobile phone PCB production workshop.