Research And Application Of Single-Stage Multi-Product Batch Scheduling Based On Quantum Genetic Algorithm

Single-stage multi-product batch process is one kind of common production lines inchemical industry. However, the scheduling of single-stage multi-product batch process is muchmore difficult than the others because of its characteristics like equipment handling flexibility,limited batch quantity and noticeable switching cost. During the past two decades, the short-termscheduling model has attracted widespread attention and research. As a new intelligenceoptimization algorithm, quantum genetic algorithm (QGA) is gradually becoming a hot spot, dueto its advantage of population diversity, being good at exploring and being easy to mix withother algorithms. This paper presents the relevant studies focused on the application of QGA inthe single-stage multi-product batch scheduling. The main work is as follows:(1) An improved QGA (Real-Coded Quantum Genetic Algorithm, RQGA) is proposed toagainst the defects of long-time convergence and easily falling into quantum"length disaster"inthe basic QGA. Firstly, an approximation operator is constructed to replace the traditionalrevolving door. It avoids the cumbersome table look-up operation and shortens the convergencetime effectively. Secondly, omitting quantum observation and encoding the quantum directly isadopted. It overcomes the "length disaster" caused by large number of quantum variables andexpands the application scope of the quantum genetic algorithm.(2) Discussion on the application of RQGA in single-stage multi-production batch process.A mixed integer linear programming model with the goal of maximizing production profits isconstructed, and through real coding to achieve the correspondence between quantum individualand process sequencing. Finally, typical example is simulated and the results have verified thevalidity and feasibility of RQGA in single-stage multi-product batch scheduling problem.(3) For the vat scheduling problem in dyeing and printing industry, a MILP model based onthe time slot ideas is established. It meets the due date condition as well as the goal of minimizing production cost, and quantum genetic algorithm is adopted to solve it. In the solutionprocess, a new order-consolidating and order-splitting method is used. The method is thatconsolidates and splits orders directly after counting the sum of the same products notconsidering the source of the orders, so the solution process is simplified. Finally, simulationresults indicate validity and feasibility of the model, as well as show the practicability ofquantum genetic algorithm in vat scheduling.Finally, a summary is made and some future works are presented.