Multiple Fault Diagnosis for Sheet Metal Fixtures Using Designated Component Analysis

摘要: This paper presents a new approach to multiple fault diagnosis for sheet metal fixtures using designated component analysis (DCA). DCAfirst defines set of patterns based on product/process information, then finds the significance these from measurement data and maps them particular faults. Existing diagnostics methods has been mainly developed rigid-body-based 3-2-1 locating scheme. Here an N-2-1 scheme is considered since parts are compliant. The proposed methodology integrates on-line data, part geometry, fixture layout sensor in detecting simultaneous multiplefixture A diagnosability discussion different type faults presented. Finally, application method presented through computer simulation. @DOI: 10.1115/1.1643076 # Fixtures used locate hold workpiece manufacturing. In general, elements can be classified by their functionality into locators clamps. Locators establish datum reference frame provide kinematic restraint. Clamps additional restraint holding position under external forces during manufacturing process. uniquely rigid body, constraining six degrees freedom part. According this principle, three placed primary plane, two secondary plane one tertiary plane. However, compliant parts, Cai et al. @1# showed that principle more adequate widely industry. establishes support it necessary than 3 due flexibility. failure directly affects location assembly dimensional quality. Ceglarek Shi @2# found launch vehicle, represent around 70% all Consequently, positively impact Due existence systems, such as optical coordinate machine ~OCMM !, large amount obtained processes. Therefore, opportunities process available. Several authors have studied last few years. past research major approaches: principal analysis, correlation clustering least square regression. 1992, Hu Wu @3# introduced ~PCA ! identify sources variation automotive body assembly. They PCA extract data. Later, @4# combining with pattern recognition. each hypothetical layout. Then, they modes production map patterns. focused single failure. Ding @5# state space model multistage processes PCA. addition, @6# include considerations noise diagnosis. Correlation able detect matching behavior measured behavior. Shiu @7# multi-station modeling critical characteristics mechanism ~fixture interactions joining conditions ~part !. expressed matrix. matrix compared simulated correlated matrices associated Multivariate also squares approach. consists relating potential causes. causes @8–12 #. Apley @9# algorithm or severity Fault was variance decomposition pattern. where determined effect element over points.