CBR系统优化策略及在注塑模具设计中的应用

CBR(Case-based reasoning)技术是弱理论强实践领域实现智能化的有效手段,但是,随着工业的飞速发展,工程领域的特征信息爆炸式增长使得当前CBR系统的关键构建环节例如事例的表达、事例库的构建等呈现出诸多局限。为了适应新的复杂工程领域发展需要,对若干关键环节进行了优化研究:在CBR系统事例的表达方面,利用GHSOM(Growing Hierarchical Self-Organizing Map)算法和粗糙集算法集成策略对事例庞大特征信息进行精简,为事例知识的准确、高效表达、存储和检索打下基础;利用GHSOM算法优化构建事例库,通过空间聚类建构多层次多粒度知识库体系,同时,当执行事例检索时,由GHSOM算法作为知识导引策略,将新任务导入相应子事例库,通过数值计算获取最佳事例。最后,以注塑模具设计为应用领域验证了本方案的效果。

OPTIMIZATION STRATEGY OF CBR SYSTEM AND ITS APPLICATION IN INJECTION MOLD DESIGN

CBR technology is an effective means to realize the intelligence of those domains, weak in theory but strong in practice. However, with rapid development in industry, characteristic information of engineering domain shows explosive growth, which makes the key links of CBR system building, such as case expression, case library construction and so on, are exposing many limitations. In order to meet the needs for the development of new complex engineering fields, this paper focuses on some key links for optimization:for case expression of CBR system, GHSOM algorithm and rough set algorithm are integrated to use in reducing the huge feature information of the case, which can lay a good foundation for correct and efficient expression, storage and retrieval of case knowledge. GHSOM algorithm is also used to optimize the construction of case library and by the spatial clustering of GHSOM, multi-level and multi-granularity of knowledge base system is constructed. At the same time, while case retrieval is executed, with GHSOM algorithm as a knowledge guidance strategy, new tasks are led into corresponding sub case library. Then, through numerical calculation the best examples are obtained. Finally, this paper takes the injection mold design as the application field to verify the effect of this scheme.