中文摘要：

装配精度预测对于产品装配精度保证具有重要意义。综合产品偏差源和偏差统一模型表达,提出偏差的旋量映射方法;在零件配合、双配合偏差传递分析的基础上,给出装配偏差有向图（Directed deviation graph,DDG）模型,并提出产品装配偏差有向图的构建流程。基于DDG模型,进一步研究用于装配精度累积的最小偏差路及其Dijkstra求解算法。在最小偏差路基础上,针对一维线性方向的装配精度预测,给出刚性零件装配偏差统计量的求解方法;针对多维空间方向的装配精度预测,提出借鉴并联机器人末端位姿计算的D-H（Denavit-hartenberg,D-H）方法,进行刚性零件装配偏差统计量的求解流程。基于装配偏差统计量,进一步给出产品装配精度评价指标,以及产品装配精度预测方法。以某产品的装配为例,介绍装配偏差统计量的求解及其装配精度预测过程。

英文摘要：

Precision predicting is important for precision confirming of mechanical assembly.Considering the deviation sources and unified representation of dimension and geometric tolerance,the twist representation for geometric deviation is discussed.By analysis of geometric deviation accumulation between two mating or double mating surfaces,the deviation accumulation model called directed deviation graph（DDG） is proposed,and its construction process is introduced.Based on the DDG,the minimum path of weight for deviation accumulation and its generation by Dijkstra algorithm is put forward,and the linear assembly precision statistic and its calculation considering the minimum path is analyzed.By Denavit-hartenberg（D-H） method,the calculation process for spacial assembly precision statistic is further proposed.Considering the assembly precision statistic,the assembly precision evaluation and predicting based on multivariate statistical analysis methodology is introduced,and an example is illustrated to verify the validity and probability of each algorithm and model.