Bottleneck Station Scheduling with Simulation Trees and Feeding Times
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current production line scheduling methods, such as those using Gurobi linear scheduling and Monte Carlo Tree Search (MCTS) combined with variable neighborhood descent, face challenges in optimizing line switching times and delivery numbers, especially at bottleneck stations with insufficient manufacturing resources, leading to increased computation and time requirements for scheduling multiple jobs.
Innovation Solution
A production line scheduling method that employs a schedule simulation algorithm to establish schedule simulation trees, calculating expected feeding times at bottleneck stations by performing node expansion based on job conditions, and using Monte Carlo Tree Search (MCTS) to generate scheduling routes, thereby optimizing job scheduling and reducing computation time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current scheduling technologies (Gurobi linear scheduling, MCTS with variable neighborhood descent) are used to optimize production schedules, then the production schedule smoothness is improved, but the computation time and computational resources required increase significantly
Solution Approach 1:
The patent performs preliminary simulation to obtain expected feeding times for all jobs at the bottleneck station before generating the final schedule. This preliminary action allows the scheduling algorithm to work with pre-calculated timing data, reducing the computational complexity during the actual scheduling phase and avoiding the need for extensive simulation of all possible schedule arrangements.
Solution Approach 2:
The patent creates a simulated model of the production line with a bottleneck station that replicates the essential characteristics of the actual production system. This simulation model allows for efficient experimentation and optimization without requiring extensive computation on the real system, enabling rapid evaluation of different scheduling approaches.
2Measurement precision
If all possible schedule arrangements are simulated to obtain objective results, then the scheduling accuracy is improved, but the computation time and resources required increase exponentially
Solution Approach 1:
The patent extracts and focuses computation on the critical bottleneck station rather than simulating all possible arrangements across the entire production line. By identifying and concentrating resources on the constraint point, the system achieves meaningful scheduling accuracy without the exponential computational cost of exhaustive simulation of all jobs at all stations.
Solution Approach 2:
The patent performs simulation only for the jobs that pass through the bottleneck station rather than simulating all possible schedule arrangements for the entire production line. This partial action provides sufficient accuracy for the critical constraint while avoiding the excessive computation required for complete exhaustive simulation.
3Adaptability or versatility
If more manufacturing machines are added to bottleneck stations to improve scheduling flexibility, then the adaptability is improved, but the cost increases due to expensive equipment
Solution Approach 1:
The patent implements dynamic scheduling that adapts to the actual state of the bottleneck station in real-time. The system continuously monitors job progress and dynamically adjusts the schedule based on current conditions, providing scheduling flexibility without requiring additional physical machines. The algorithm can resequence jobs and adjust timing to accommodate variations in processing speed and availability.
Solution Approach 2:
The patent changes the parameters of existing jobs and processes to optimize bottleneck utilization. By adjusting job sequencing, timing parameters, and feeding rates, the system achieves improved adaptability and throughput using the same physical resources, avoiding the need to add expensive new manufacturing machines.
Data Source
AI summary
A production line scheduling method, adapted to a plurality of jobs passing a bottleneck station having at least one manufacturing machine, the jobs respectively correspond to a plurality of job conditions, and the method includes: performing a plurality of times of a schedule simulation algorithm on the jobs to sequentially establish a plurality of schedule simulation trees, and obtaining a job schedule and a simulated finishing period of each job based on the schedule simulation trees; and calculating a plurality of expected feeding times of each job at a plurality of stations including the bottleneck station, each schedule simulation tree includes at least one scheduling route, and each scheduling route is generated from one schedule simulation algorithm, the schedule simulation algorithm includes: performing a node expansion step based on at least one node expansion condition and the job conditions to obtain the scheduling route.