An improved case-based reasoning approach for mechanical design by enhancing the retrieval accuracy and leveraging the implicit knowledge

Case-based reasoning is extensively applied in mechanical parametric design to improve automation. However, solving new problems accurately using the implicit knowledge in existing mechanical product cases remains challenging. This study proposes an improved case-based reasoning (ICBR) method designed to effectively retrieve parameters of varying weights in the database and extract the implicit knowledge from the problem and solution of candidate cases, thereby enhancing the precision of the design. The content of this study consists of two parts. The first part proposes a case retrieval method using a novel hybrid weight (CRNHW), which merges subjective and objective weights to prioritize different design parameters. This method identifies candidate cases, encompassing both problems and their corresponding solutions. The second part introduces a case adaptation approach using a hybrid weighted mean (CAHWM). This approach decouples implicit knowledge and covers four aspects: the similarity of new and candidate case problems (S), the utility of each candidate case (U), the gray relationship between each candidate case problem and solution (R), and the feature adaptability of each solution (A). We can derive the solution by integrating these four types of implicit knowledge. Finally, we adopt ICBR in the parametric design of actual mechanical presses to illustrate its feasibility and carry out comparison experiments with different classical methods. The experimental results prove that ICBR performs better with 84.49% adaptation accuracy when the number of candidate cases is 10.