基于改进D-S证据理论的滚抛磨块融合决策模型

目的 为了能够有效利用滚磨光整加工数据库平台的案例知识和专家经验,提高新零件加工时滚抛磨块优选的准确率,解决不同优选方式优选结果的冲突问题.方法 将案例推理、专家推理、迁移学习3种优选方式的滚抛磨块优选结果作为3种证据,根据3种优选方式计算的相似度结果构建滚抛磨块决策辨识框架,并采用合理的方法确定基本概率赋值.然后依据...

Fusion Decision Model of Tumbling Chip Abrasives Based on Improved D-S Evidence Theory

As the primary solid medium in the barrel finishing process, the tumbling chip abrasives has a great influence on the processing effect. In order to effectively utilize the case knowledge and expert experience of the database of barrel finishing process, and improve the accuracy of the optimization of tumbling chip abrasives when the new part are processed, the research group have established the optimization model of tumbling chip abrasives based on case-based reasoning, expert reasoning and transfer learning respectively. However, optimization result only based on three independent methods had low reliability,for the new part to be processed, there will be conflicts among the three optimization results, so that it is necessary to make a fusion decision for the three optimization results. Therefore, a fusion decision model of tumbling chip abrasives based on improved D-S evidence theory is proposed. Firstly, the optimization results of case-based reasoning, expert reasoning and transfer learning are used as three kinds of evidence. According to the similarity results calculated by the three optimization methods, the decision frame of discernment of tumbling chip abrasives is constructed, and a reasonable method is used to determine the basic probability assignment. Secondly, aiming at the problem that the fusion results are contrary to the actual situation when the evidences are highly conflicting in the traditional D-S evidence theory, the method of distributing the basic probability assignment according to the proportion of conflict information is used to improve the synthesize formula. Then, the improved synthesize formula is used to fuse the three kinds of evidence. Finally, the simulation is carried out by using the real data of factory processing in the database. Based on existing case results, a large number of simulation results show that the improved fusion decision model can solve the conflicts between the optimization results of different optimization methods as well as the disadvantages of the original synthesis formula. The results of fusion decision have higher accuracy than those of the other three methods. The accuracy of the fusion decision model reaches 88%, which shows that the proposed decision model can provide decision guidance for intelligent optimization of tumbling chip abrasives.