Product complexity increases along with the increase of product functions. Effective methods are required in the complex product development. Axiomatic design can effectively reduce product complexity, but there are deficiencies in the usage process and the innovation stimulation. This paper proposes a holistic method for the development of complex products by using a neural network-aided technological evolution process to control negative effects of complexity. For the first time, engineering parameters are used as the intermediary between the complexity and technological evolution. Problematic engineering parameters are extracted from a current reality tree model through a natural language process. An artificial neural network is trained to predict the technological evolution law based on existing successful samples. Appropriate analogical objects are formed in an invention principle case library through the similarity analysis of engineering parameters. The optimal scheme is developed objectively by using the coefficient variation method and Dempster combination rule. The proposed method is applied to develop a pipe cutting machine with a granted patent for its feasibility and effectiveness.

产品复杂度随着产品功能的增加而增加。在复杂的产品开发中需要有效的方法。公理化设计可以有效地降低产品复杂性，但是在使用过程和创新刺激方面存在不足。本文提出了一种通过使用神经网络辅助的技术演化过程来控制复杂性的负面影响的开发复杂产品的整体方法。首次将工程参数用作复杂性和技术发展之间的中​​介。通过自然语言过程从当前的现实树模型中提取有问题的工程参数。训练了一个人工神经网络，可以根据现有的成功样本来预测技术发展规律。通过对工程参数的相似性分析，在发明原理案例库中形成适当的类比对象。利用系数变化法和Dempster组合法则，客观地确定了最优方案。所提出的方法被用于开发具有可行性和有效性的已获专利的切管机。