A geometric similarity case-based reasoning system for cost estimation in net-shape manufacturing

Cost estimating is an essential process in design for both the designer and the manufacturer. From the designer's point of view, it acts as a tool for evaluating different design alternatives. It is useful for calculating the feasibility of the design compared to similar products. From the manufacturer's point of view, much of the success in winning job orders hinges on the ability of the manufacturer to provide the best (lowest) cost estimate within a short period of time. Net-shape manufacturing (NSM) processes such as die casting, injection molding, stamping, and forging are different from machining operations in the sequence of creating component features. These processes are used to produce all features of a component in one shot. Therefore, the cost estimating approaches used in machining operations based on the process plans are not appropriate for NSM.; The cost of a component produced by NSM consists mainly of startup costs and individual component cost. Startup costs are amortized over the production quantity and include tooling, engineering, layout, and prototyping costs. Costs of individual components include material cost and processing cost. Unlike machining operations, tooling and processing costs represent a major share in the component cost and depend highly on its geometric complexity. Recognizing geometric complexity of the component from its CAD model is not a straightforward task. So, a case-based reasoning (CBR) approach is suitable for these kinds of processes to quantify the geometric complexity based on historical components.; In this research, building a database of historical components together with their geometric complexities and estimated costs will be described. The representation scheme used for the database is called attributed graphs, which describes the topology of the mechanical component and incorporates some geometric information. The graph representation is based upon the CAD information exported from the CAD system in STEP format. Indexing and retrieving components in the database is based on a rough similarity criterion such as total number of surfaces. Similarity assessment between two components is performed by comparing the two graphs representing those components. Two approaches for graph comparison are reported in this dissertation.; A method for quantifying the complexity of the component manually is reported. The method is based on field information collected form manufacturers about the geometric features that add to the manufacturing complexity. This method is helpful in calculating the complexity of the initial components to be stored in the database of the CBR system. Finally, a model for cost estimation system has been developed and implemented. The system starts with the information extracted from the CAD system and builds the cost estimate based on the cost factors submitted by the company. To produce a cost estimate for a component, the system needs a database of company practices to be filled out, a database of historical components together with their complexities, and the data extracted from the CAD system.