Process parameters of injection molding are the key factors affecting the final quality and the molding efficiency of products. In the traditional automatic setting of process parameters based on case-based reasoning, only the geometric features of molds are considered, which may not be the representative feature of products and cause the reasoning process to fail. This problem of failure manifests itself in that the molding process parameters inferred by the reasoning system may be very different between molds with similar geometric features or very similar between molds with different geometric features. Therefore, this paper proposes a case-based-reasoning method based on molding features in order to overcome this problem by a method of dimensionality reduction, composed of three stages which (1) obtain the injection pressure profile data through actual injection molding or filling simulation analysis, (2) calculate the similarity of the pressure profiles between target case and each of source cases in case database using the nearest neighbor method, and sort according to the value of similarity, (3) find the case with a maximum of similarity out as the one closest to the target case, and take the process parameters of the most similar case as the solution of the target case according to case modification strategies. This method simplifies the high-dimensional molding features to the pressure profile at the injection location with two-dimensional data features. Experiments show that the new method has a high retrieval accuracy and sensitivity. Moreover, even slight differences in molding can be captured easily.

注塑成型的工艺参数是影响产品最终质量和成型效率的关键因素。在基于案例推理的传统工艺参数自动设置中，仅考虑模具的几何特征，这可能不是产品的代表性特征，并且会导致推理过程失败。失败的问题本身表现为，由推理系统推断的成型工艺参数在具有相似几何特征的模具之间可能非常不同，或者在具有不同几何特征的模具之间可能非常相似。因此，本文提出了一种基于造型特征的基于案例的推理方法，以通过降维方法克服这一问题，由三个阶段组成：（1）通过实际的注塑成型或填充模拟分析获得注射压力曲线数据，（2）使用最近邻方法在目标数据库中计算目标案例与每个源案例之间的压力曲线相似度，并根据相似度的值进行排序，（3）找出最大相似度的案例作为最接近目标案例的案例，并根据案例将最相似案例的处理参数作为目标案例的解。修改策略。该方法使用二维数据特征将高维成型特征简化为注射位置处的压力曲线。实验表明，该方法具有较高的检索精度和灵敏度。此外，