基于集成学习算法的可持续模块划分方法

Sustainable Module Partition Method Based on Integrate Learning Algorithm

随着客户对可持续产品需求的增加,针对产品功能和结构属性的传统模块化设计,正逐步转向可持续模块化设计。然而当前的模块划分方法更注重于对模块化指数的优化迭代,忽略了产品多组件间的信息传递和反馈。为此,提出一种集成学习算法的可持续模块划分方法。首先,面向组件间的回收性、材料和寿命因素,构建产品的综合DSM模型。其次,提出集成学习算法的弱分类器构建规则和强分类器结合策略。最后,通过颚式破碎机案例验证了所提模块划分方法的可行性,集成学习算法同直接聚类和遗传算法的模块划分结果对比,表明所提方法各模块组件更接近DSM的对角线。

With the increase of customer demand for sustainable products,modular design for function and structural attributes is gradually shifting to sustainable modular design.However,the current method of module partition pays more attention to the optimization iteration of modularity index,ignoring the information transfer and feedback between multiple components.Therefore,a sustainable module partition method based on integrated learning algorithm was proposed.Firstly,a comprehensive DSM model of the product was constructed for recyclability,material similarity and service life of the components.Secondly,the construction rules of the weak classifier and the combination strategy of the strong classifier were proposed based on integrated learning algorithm.Finally,a jaw crusher case was used to verify the feasibility of the proposed method.Compared with direct clustering and genetic algorithm,the results of ensemble learning algorithm are closer to the diagonals of DSM.