A Hybrid Expert System for Estimation of the Manufacturability of a Notional Design

Figure 3

Logical pipeline of our system. Each subtree (–) figuratively represents the variables and rules concerning a particular cost criterion. The user supplies the inputs to both the ES (via the input variables, here represented as red and dark blue circles) and the weights used to modify the aggregation of the criteria scores . From some user-defined set of weight arrays (ex: array A, array Z) the user can select those which are used to participate in the weighted sum (summation at the right of the pipeline) which produces the manufacturability score (M). Dark blue circles represent control variables, which decide which input variables are active. Red circles represent input variables supporting rule execution. Light blue circles represent intermediate values produced by rule execution. Blue squares are penultimate variables used to compute the criteria subscores. Mathematically, these operations are executed in one of three ways. Fuzzy logic is used to convert crisp input values into fuzzy intermediate values in a process called fuzzification. Fuzzy rules map between fuzzy values. Last are defuzzification operations, which are not rules. The concluded weighted average uses predefined weights defined by the experts. More information is available in Section 3.2.3.