Risk assessment plays a critical role in the product design, manufacture and production process for different engineering fields. Traditionally, risk assessment-related issues are addressed using the failure mode and effects analysis approach to perform reliability analysis. However, information provided by experts in the risk assessment process often includes complete information, incomplete information and hesitant information at the same time. The traditional failure mode and effects analysis approach cannot handle incomplete information and hesitant information in the risk assessment process. Moreover, the traditional failure mode and effects analysis approach does not take into account the subjective and objective weights of the risk factors. In order to effectively overcome the limitations of the traditional failure mode and effects analysis method, this paper proposes a novel flexible risk assessment approach integrating subjective and objective weights under uncertainty. For handling the incomplete information, the proposed method uses currently known information to fill in the missing information. On the other hand, the proposed method uses hesitant fuzzy linguistic term set instead of single linguistic term set to handle hesitant information in the risk assessment process. Moreover, the proposed uses the concept of statistical variance to calculate the objective weights of the risk factors. An illustrative example of the screen printing stage of photovoltaic cell manufacturing is adopted to demonstrate the rationality and practicability of the proposed method. This paper also compares the risk priority results of the proposed method with those of the traditional FMEA method, the hesitant fuzzy linguistic term set method, and the Rao and Patel (2010) method. The simulation results indicate that the proposed can provide more correctly rank the identified causes of failure. The proposed not only simultaneously considers the subjective and objective weights for the risk factors, but also can handle hesitant information and missing information of the assessment attribute values.

风险评估在不同工程领域的产品设计，制造和生产过程中起着至关重要的作用。传统上，与风险评估相关的问题是使用故障模式和影响分析方法来执行可靠性分析的。但是，专家在风险评估过程中提供的信息通常同时包括完整信息，不完整信息和犹豫信息。传统的故障模式和影响分析方法无法在风险评估过程中处理不完整的信息和犹豫不决的信息。而且，传统的故障模式和影响分析方法没有考虑风险因素的主观和客观权重。为了有效克服传统失效模式和影响分析方法的局限性，提出了一种将不确定性下的主客观权重相结合的新型柔性风险评估方法。为了处理不完整的信息，所提出的方法使用当前已知的信息来填充丢失的信息。另一方面，提出的方法使用犹豫模糊语言术语集代替单一语言术语集来处理风险评估过程中的犹豫信息。此外，该提议使用统计方差的概念来计算风险因素的客观权重。以光伏电池丝网印刷阶段为例，说明该方法的合理性和实用性。本文还比较了该方法与传统FMEA方法，犹豫模糊语言术语集方法以及Rao和Patel（2010）方法的风险优先级结果。仿真结果表明，所提出的算法可以对识别出的故障原因进行更正确的排序。提出的方法不仅可以同时考虑风险因素的主观和客观权重，还可以处理评估属性值的犹豫信息和缺失信息。