The Research Andapplication Of Fuzzy Job-Shop Scheduling Problem

With the arrival of the global economy integration,the competition in world market is becoming more and more fierce.In order to keep the unbeaten status,enterprises should improve their production management.Production scheduling has gradually been paid attention as the core of the production management, on which the relationships are researched among resources, job, time and performance index. The Job-Shop Scheduling Problem (JSP) is the most classical and common in this field.JSP has been proved as a NP-hard problem,which is complicated and multi-objective.For a long time,various intelligent computation methods have been used to solve JSP by Scholars.Genetic Algorithm is widely used in JSP because of its less-dependency and robustness.Now, most of the studies on the JSP are static; but in the actual production process,the processing time and delivery time are uncertain because of random factors.So in this paper,we have researched the problem in deeply based on current theoretical basis.At first , this paper expounds the production schedule problem and its developing state,the definition of JSP has also been given.On this basis,the mathematical model of JSP is studied.In order to overcome the disadvantages of prematurity and slow convergence in genetic algorithm,a new Improved Genetic Algorithm based on sequence-based code and Two Points Order Crossover(TPOX) was designed.This new algorithm avoids generating the illegal chromosomes and accelerates the convergence speed of the optical value.According to the test results in solving the eight classical Benchmarks problem,proving the effectiveness of the new algorithm.Secondly,on the basis of fuzzy theory,we use of triangular fuzzy number for the fuzzy processing time and trapezoidal fuzzy number for the fuzzy due-date. We also defined the mathematical model of the fuzzy JSP problem in three different situation. We designed a genetic algorithm in which generate the initial population with the conception of similarity, contrary to the objective function of maximum weighted satisfactoriness based on the fuzzy processing time and fuzzy due-date. In the meantime,the method of calculating the value of fitness is discussed in deeply.At last,we used the classical Sakawa problems to checking the genetic algorithm for the fuzzy JSP, and simulated a scheduling example based on Hong-Xia machinery plant. The analysis of simulation result shows that the improved genetic algorithm is effective and feasible in solving the fuzzy JSP problem. Furthermore the improved genetic algorithm can guide the actual job-shop production and it is worth popularizing.