基于客户满意度的模糊神经网络JSP研究

针对具有不确定性因素的作业车间调度问题，基于模糊数学的思想，把模糊加工时间、间隔期和模糊交货期用梯形模糊数表示，建立了基于客户满意度曲模糊作业车间调模型。运用Hopfleld神经网络算法求解，结合目标函数和JSP的全部约束条件，构建能量函教和JSP换位矩阵，保证了神经网络稳态输出为最优生产调度方案。最后用网络计划图对稳态输出的换位矩阵进行解码得到最优调度甘特图，避免了传统成本树法易出现死锁调度的问题。计算实例验证了本算法的可行性和有效性。

Study on Fuzzy Neural Network JSP Based on Customer Satisfaction

For the job-shop scheduling with the uncertain factors, and according to the fuzzy theory, Fuzzy processing tinge,interval and fuzzy due date were denoted by trapezoid fuzzy number. A fuzzy job shop scheduling model was built up based on clients＇ degree of satisfaction. Hopfield neural network algorithm was used to solve the problem. With the objective function and constraints of JSP,energy function and JSP commutants were given, in order to ensure that steady-state output of neural network was the best scheduling options. At last, the network plan graph was used to decode steady-slate output matrix transpesition as optimal scheduling Gantt chart. In this process, Deadlock scheduling problems were avoided ,which often appeared in traditional cost tree method. The calculating example shows that the algorithm is feasible and effective.