Domain knowledge based non-linear assembly sequence planning for furniture products

Abstract Adopting assembly sequences that meet the design and assembly requirements of a product is an effective way to make the product assembled effectively and satisfy its design requirements accurately. However, generating assembly sequences that have the above characteristics is still very difficult, and effective & automatic approaches are still rare. In this paper, furniture products are taken as research objects. To enable an input furniture model (digital assembly model of a furniture product) to smartly obtain its assembly sequences having the above characteristics based on its implicit design requirements (function & its realization manner and geometry), a novel automatic non-linear assembly sequence planning approach is presented. The criteria for determining whether a sub-assembly of furniture models is a functional sub-assembly (meeting the design and assembly requirements of furniture products) are defined first. Meanwhile, an attributed part layout graph is adopted to represent the input model at a semantic level. Then, based on the criteria and the above graph, each functional sub-assembly in the input model will be accurately and gradually recognized by adopting a heuristic graph searching method; matching along with this process, a layout-based hierarchical graph is also gradually generated to effectively represent all the recognized functional sub-assemblies and their relationships by using a graph clustering method. Finally, an improved particle swarm optimization method, integrating with a heuristic algorithm, is adopted to hierarchically generate the linear high quality assembly sequences for the input model based on the above hierarchical graph. This step also brings the final non-linear high quality assembly sequences of the model, which are stored by using an extended partial assembly tree. Experimental results are presented to demonstrate the effectiveness of the approach.