On-line predictive model for disassembly process planning adaptation

Abstract

This paper presents an on-line predictive model for disassembly process adaptation. The prediction enables a planner to adapt the process plan based on the condition of the product (e.g., degree of rustiness, deformation) during process execution. This model tries to correlate the product physical condition, used as an explanatory variable, with the component value and disassembly cost, the response variables. The core of the approach is based on an inference engine that used a kernel regression. A simple methodology for integrating the predictive planner in a disassembly system is presented and exemplified by a case study of the disassembly of a ratio.