LAUSR

Abstract The mechanical deformation of workpieces due to tightening is a common phenomenon in most assembly processes. Such deformation is typically characterized on the basis of a few critical points from sensing signals during process monitoring. Our previous study focus on improving critical point detection accuracy by establishing a state space model and a two-stage particle filter algorithm. The state variables are estimated in the first stage and the critical point is estimated in the second stage. These two stages are recursively estimated until the estimation of critical point converges. However, such method usually requires a large amount of computational efforts which may not be affordable in practice. To effectively identify critical points as well as meet the timeliness of detection, we improve the estimation algorithm by leveraging the quantification of state changes and estimating the critical point in the first stage. In this way, the critical point can be identified within one stage, thereby significantly reducing computation costs. The results from a real case study indicate that our proposed method delivers efficient critical point detection performance for process monitoring.