Abstract: In the existing manufacturing readiness level assessment of complex product assembly, the index weight and index score were evaluated by experts from experience. This resulted in some deficiencies such as subjectivity, heavy work, long time, and non-impartment of knowledge in the assessment cases. To improve the efficiency and objectivity of manufacturing readiness level assessment of complex product assembly, utilizing the dataset of manufacturing readiness level assessment cases, the manufacturing readiness level assessment was discussed herein based on BP artificial neural network and AdaBoost algorithm. A manufacturing readiness assessment index system of complex product assembly was established. The quantification of index and readiness level assessment were proposed based on fuzzy evaluation and membership function. Then the manufacturing readiness level assessment of complex product assembly was modeled based on BP neural network. The AdaBoost algorithm was applied to optimize readiness level assessment model based on BP neural network. To optimize the assessment model, it is trained on the dataset of manufacturing readiness level assessment cases and the results of BP-AdaBoost algorithm was analyzed. The optimal assessment model was obtained. Experimental results show that the assessment is good in reliability and accuracy based on BP-AdaBoost algorithm.

Key words:  , product assembly, manufacturing readiness, level assessment, BP artificial neural network, AdaBoost algorithm

摘要： 现有复杂产品装配制造成熟度等级评估依赖专家凭经验确定指标权重和指标评分，存在主观性较强、工作量大、耗时长、无法传承评价实例所蕴含的知识等问题。为了提高复杂产品装配制造成熟度等级评估的效率以及客观性，利用成熟度等级评价实例数据，研究基于BP人工神经网络和AdaBoost算法的制造成熟度等级评估方法。构建复杂产品装配制造成熟度评价指标体系，给出基于模糊评价法和隶属函数的评价指标及成熟度等级达成度量化方法，建立基于BP神经网络的复杂产品装配制造成熟度等级评估模型，并使用AdaBoost算法优化成熟度等级评估BP神经网络模型。采用复杂产品分系统装配制造成熟度评价数据集对评估模型进行训练和实验，分析BP-AdaBoost的评估结果，获得最优评价模型。实验结果表明，基于BP-AdaBoost算法的复杂产品装配制造成熟度等级评估方法具有较好的可靠性与准确度。

关键词: 产品装配, 制造成熟度, 等级评估, BP神经网络, AdaBoost算法