Improved Jacobian-Torsor model for optimizing mechanical product quality

Abstract

The quality of an assembly largely depends on the predictive accuracy of the assembly deviation model in the selective assembly process. Presently, three-dimensional deviation analysis methods only offer tolerances that indicate the overall deviation of a part feature. This limitation fails to accurately represent the fit of the parts during the assembly process, resulting in initial inaccuracies in the selective assembly model. To address this issue, an improved Jacobian-Torsor model is proposed that incorporates measured data from part assembly features into the deviation analysis model. Initially, the actual contact state of the assembly is determined by measuring the contact surface of the plane fit and the shaft-hole fit. Next, the Jacobian matrix, representing the transmission path of assembly deviation, is constructed by analyzing the assembly connection diagram. Subsequently, by integrating the actual contact state characterized by small displacement torsor (SDT) into the unified Jacobian-Torsor model, the advantage of high credibility tolerance propagation is provided. Finally, the feasibility of the proposed improved Jacobian-Torsor model is verified using the double-joint manipulator.