Reliability improvement on assembly accuracy with maximum out-of-tolerance probability analysis and prior precise repair optimization

Abstract

For complex thin-walled structures, the accumulation of different error sources behaves a complex nonlinear relationship, causing the frequent occurrence of out-of-tolerance phenomenon, and retention ability of assembly accuracy is weak. Novel methods for improving accuracy reliability with probability analysis and prior precise repair compensation before practical assembly operations were studied. Firstly, considering the non-independent relationship among multi-dimensional error sources, their internal stress function with practical error items and the EOCDCF (Extreme Out-of-tolerance Caused by Deviation’s Coupling&Focusing Function) phenomenon were analyzed. Secondly, the probability method was adopted for evaluating and improving accuracy reliability. To be more specific, the occurrence probability model of EOCDCF event was established with Gartner-Ellis theorem in LDP (Large Deviation Principle), and the dynamic sliding window method was adopted to calculate its probability. Although feedback adjustment on reliability results could be proposed to avoid EOCDCF event considering online information, once it occurred, to control assembly accuracy and internal stress, the accurate repair simulation method based on IPSO (Improved Particle Swarm Optimization) algorithm was proposed considering the actual deviations, and detailed repair areas and quantity on parts would be designed in advance. Finally, a typical wing-box component was verified, and its assembly quality improvement rate was about 30%. This solution could enhance accuracy reliability and ensure assembly accuracy within the required range, and would also lay the precision foundation for product’s service performance.