Smart manufacturing is an essential part of fourth industrial revolution in which robotic machines can control and perceive automatically to provide effectiveness and convenience in production process. However, the existence of potential failures and defects not only influence the manufacturing process but also damages the resources and cause negative impacts on environment. Failure modes and effects analysis (FMEA) is a key approach to identify and eliminate possible failures and evaluate the risks from design, system and process. This research paper provides a novel FMEA approach for risk evaluation by integrating rough set theory and ELimination and Choice Translating REality (ELECTRE) II method to handle the subjectivity and uncertainty in experts’ judgements without much prior information, membership functions and additional adjustments. Rough numbers are used to study uncertainty in linguistic terms using intervals instead of single fixed values. The proposed approach is formulated by defining different types of concordance and discordance sets using optimization techniques based on statistical dispersion and maximum deviation method. The presented technique shows the strong, weak and neutral pairwise relations among failure modes by systemically comparing them from each risk component. The distance functions and averaging methods are applied to check the similarities and differences among error modes which improves the accuracy of the results. The developed rough FMEA approach is applied to identify the potential failures of robot working in optical cable industry and evaluate the risk components of manufacturing and production process. Rough ELECTRE II approach can be effectively applied to enhance the efficiency of working conditions and prevent the loss of crude materials and energy.

智能制造是第四次工业革命的重要组成部分，在第四次工业革命中，机器人机器可以进行控制和自动感知，以在生产过程中提供有效性和便利性。然而，潜在的故障和缺陷的存在不仅影响制造过程，而且还会破坏资源并对环境造成负面影响。故障模式和影响分析（FMEA）是识别和消除可能的故障并评估设计，系统和过程中的风险的关键方法。本研究论文通过将粗糙集理论与消除和选择转换现实（ELECTRE）II方法相结合，提供了一种新颖的FMEA风险评估方法，无需过多的先验信息，隶属函数和其他调整，即可处理专家判断中的主观性和不确定性。粗糙数用于使用间隔而不是单个固定值来研究语言术语中的不确定性。通过使用基于统计分散和最大偏差法的优化技术定义不同类型的一致性和不一致集来制定提出的方法。提出的技术通过系统地比较每个风险组件中的故障模式，显示了故障模式之间的强，弱和中性成对关系。距离函数和平均方法用于检查误差模式之间的异同，从而提高了结果的准确性。所开发的粗略FMEA方法用于识别在光缆行业中工作的机器人的潜在故障，并评估制造和生产过程中的风险成分。