| Summary: | 碩士 === 國立臺灣科技大學 === 工業管理系 === 104 === Recently, the way to raise benefit to any company is not only increasing sales and
decreasing cost, but improving job scheduling efficiency in the extremely competitive
environment. Therefore, it is necessary for enterprise to provide efficient scheduling plan to
achieve quick response. Especially in the era of Industry 4.0, it is more important to implement
the Intelligent Scheduling System.
Production scheduling problem, mainly, is used to distribute resource efficiently to raise
up production efficiency, to cost down, and to shorten machine idle time while reducing total
working time. Optimizing production scheduling can decrease enterprise operational cost,
production development time and manufacturing time to reach the enterprise goal which are
minimizing the total cost and quick response to customer. Thus, it will be more competitive for
enterprise in market.
This thesis is focus on the Job Shop Scheduling Problem (JSSP), which is an extension
and more complex than the Flow Shop Scheduling Problem. To solve the JSSP, this thesis
proposes a two-stage algorithm, called improved Ant Colony Optimization (iACO), based on
the Ant Colony Optimization Algorithm (ACO) and Roulette Wheel Selection (RWS). In the
iACO, the RWS would choose one from all of the pheromone rule and then schedule for the
JSSP according to rule chose in the RWS.
In this thesis, we implement the iACO by java and test its efficiency and elasticity by 82
job shop benchmark problem from the OR-library. These benchmark problems have different
number of machines and jobs ranging from 66~5010. The iACO algorithm provided in this
thesis find superior and more stable solution than those traditional scheduling methods, such
as Earliest Due Date, First Come First Served, Longest Processing Time, Shortest Processing
Time, Weighted Shortest Processing Time and Critical Ratio.
|
|---|
