JAPAN/USA Symposium on Flexible Automation 2000 MODELING AND DIAGNOSIS OF MULTISTAGE MANUFACTURING PROCESSES: PART I – STATE SPACE MODEL

This two-part paper focuses on the modeling and diagnosis of multistage manufacturing processes. Part I develops a state space model characterizing variation propagation in a multistage process. The state space model describes a discrete-time LTV (Linear Time Varying) stochastic system, which strongly indicates that the existing system and control theory could be used to perform systematic analysis and achieve variation control of the manufacturing process. Moreover, the state space model integrates information of product quality with process information such as tooling status, therefore providing the basis for fault diagnosis. The model is validated through simulation comparison with the widely used software, Variation System Analysis (VSA). The work of fault diagnosis in a multistage manufacturing process is continued in Part II. Despite the research is presented in the specific context of sheet metal assembly process, the methodology is applicable to generic multistage manufacturing processes. NOMENCLATURE: J, N the index and number of all parts in final product MLPξ the ξ Measurement Location Point at part J PLP Primary Locating Point STD standard deviation U(i) fixturing deviation at station i, defined as [ ]T 2 n 1 n 12 11 ) i ( P ) i ( P ) i ( P ) i ( P i i ∆ ∆ ∆ ∆ ) i ( J X the state variable that is the deviation vector of part J at station i, defined as [ ] T J J J ) i ( ) i ( Z ) i ( X α ∆ ∆ ∆ X, Z the global coordinate variables i, m the index and number of stations s, ni, the index and number of subassemblies at station i x, z the local coordinate variables Y y, deviation vector of measurement point ∆ deviation operator λ the index of the station on which part J is at its first time welded with other parts or subassemblies in the assembly stream