Elsevier

Applied Soft Computing

Volume 49, December 2016, Pages 734-747
Applied Soft Computing

A perceptual computing-based method to prioritize failure modes in failure mode and effect analysis and its application to edible bird nest farming

https://doi.org/10.1016/j.asoc.2016.08.043Get rights and content

Highlights

  • A new RPN model based on Perceptual Computing is proposed.

  • Linguistic uncertainties pertaining to group decision making in FMEA are considered.

  • Providing RPN scores in both numerals and linguistic words.

  • A better insight pertaining to the risk of a failure mode is provided.

  • A case study related to edible bird nest farming in Borneo Island is reported.

Abstract

A failure mode and effect analysis (FMEA) procedure that incorporates a novel Perceptual Computing (Per-C)–based Risk Priority Number (RPN) model is proposed in this paper. The proposed model considers linguistic uncertainties and vagueness of words, because it is more natural to use words, instead of numerals, for an FMEA user to express his/her knowledge when he/she provides an assessment. Therefore, it is important to consider the inherited uncertainties in words used by humans for assessment as an additional risk factor in the entire FMEA reasoning process. As such, we propose to use Per-C to analyze the uncertainties in words provided by different FMEA users. There are three potential sources of risks. Firstly, the risk factors of Severity (S), Occurrence (O), and Detection (D) are graded using words by each FMEA user, and indicated as interval type-2 fuzzy sets (IT2FSs). Secondly, the relative importance of S, O, and D are reflected by the weights given by each FMEA user in words, which are indicated as IT2FSs. Thirdly, the expertise level of each FMEA user is reflected by words, which are expressed as IT2FSs too. The proposed Per-C-RPN model allows these three sources of risks from each FMEA user to be considered and combined in terms of IT2FSs. A case study related to edible bird nest farming in Borneo Island is reported. The results indicate the effectiveness of the proposed model. In summary, this paper contributes to a new Per-C-RPN model that utilizes imprecise assessment grades pertaining to group decision making in FMEA.

Introduction

Failure Mode and Effect Analysis (FMEA) is an effective methodology for determining the postulated component failures/errors of a process, system, or design [1]. In manufacturing processes, FMEA is useful to define, identify, and reduce the potential failures of a process prior to their occurrence by eliminating their root causes, thereby improving safety, reliability, and quality of the operations [1], [2]. Among the successful applications of FMEA include automotive [1], aerospace [3], agriculture [4], chemical [5], health care and hospital [6], [7], manufacturing [8], mechanical [9], nuclear [10], electronic and semiconductor [11] and ocean engineering [12]. Traditionally, the Risk Priority Number (RPN) is used to determine the risk associated with a failure. To obtain the RPN score, one can simply multiply three risk factors, i.e., Severity (S), Occurrence (O), and Detection (D). As such, RPN = S × O × D [2], where S and O are the seriousness and frequency of a failure mode, respectively, while D is the effectiveness of the existing measures in detecting a failure before it affects the customers [1]. While the conventional RPN model is straightforward, it has a number of limitations [2], [13], [14], [15], [16], [17]: (i) the RPN score does not consider the relative importance pertaining to S, O, and D; (ii) while the same RPN score can be produced by different combinations of S, O, and D, the underlying risk implication can be different; (iii) it is not easy to precisely assess S, O, and D; (iv) the method for computing the RPN score (i.e., by multiplying S, O, and D) is open to discussion, as it is not robust in terms of evaluating critical factors.

Many solutions have been proposed to tackle the shortcomings of the conventional RPN model [2]. A search in the literature reveals that two popular research trends pertaining to FMEA, which are also the focus of this paper, are: (i) imprecise assessment grades [13], [14], [15], [16], [17], and (ii) group decision behaviours [2], [13], [14], [15], [16], [17], [18]. The first trend suggests the use of uncertain, imprecise, and vague words for assessing S, O, and D, because it is more natural for one to express his/her knowledge in such way [13], [14], [15], [16], [17], [19]. The second trend suggests that FMEA is a cross-functional team activity, and different FMEA users provide different opinions because of varying expertise and background [2], [13], [14], [15], [16], [17], [19]. A challenge in the abovementioned two research trends is the possibility of linguistic uncertainty among FMEA users. In this paper, it is argued that the inherent uncertainties in words have to be carefully handled in order to avoid unnecessary loss of information as well as to enhance the validity and applicability in risk analysis. This challenge often occurs in FMEA activity [17], [20]. Therefore, the aim of this paper focuses on utilizing IT2FS representation and the relevant techniques for tackling the abovementioned challenge.

With reference to fuzzy set theory, intra- and inter- uncertainties are two existing uncertainties in using words to express one’s knowledge [20], [21], [22], [23], [24], [25], [26], [27]. The former is concerned with the uncertainty of words used by an individual, while the latter relates to the uncertainty of words used by a group of individuals [20], [21], [22], [23], [24], [25], [26], [27]. In this paper, we argue that handling both uncertainties is important because different FMEA users express their opinions in different ways in accomplishing a team-based activity, viz. intra- and inter- uncertainties exist simultaneously, and they imply risks [22], [23].

A relatively new line of study suggests that T1FSs are not suitable to handle both uncertainties owing to a number of reasons [22], [23], [24], [25], [26]: (i) an T1FS with all its parameters defined implies no uncertainty for the word used; (ii) different people may interpret the same word differently; therefore, uncertainties exist; and (iii) it is contradicting to use a “certain” (or precise) model to represent uncertainties. Even though the practicalities and contributions of type-1 fuzzy methods in FMEA [13], [14], [15], [16], [17] are known, it is not clear how intra- and inter- uncertainties of words should be handled with T1FSs [20], [21], [22], [23], [24], [25], [26]. These challenges can potentially be solved by using an extension of T1FSs, i.e., interval type-2 fuzzy sets (IT2FSs) [22], [23], [24], [25], [26], [27], [28]. While IT2FSs have received much attention, the use of IT2FSs in FMEA is relatively new [2], [29]. Comparing with T1FSs, IT2FSs offer a more realistic means with the capability of modelling second-order uncertainties [30], [31], [32], [33]. A number of theoretical investigations with respect to the properties of IT2FSs have been developed [22], [23], [24], [25], [27], [30], [31], [32], [33], [34]. Many applications of IT2FSs have been reported too, e.g. [22], [23], [24], [25], [26], [28], [29].

Motivated by the usefulness and flexibility of IT2FSs, a new perceptual computing (Per-C)–based RPN (hereafter abbreviated as Per-C-RPN) model with IT2FSs is proposed. Per-C is adopted owing to its effectiveness in handling inherent uncertainties in words [23]. Specifically, Per-C is able to handle subjectivity, vagueness, imprecision, and uncertainty while achieving tractability and robustness in modelling human decision-making behaviours [22], [23], [24], [25], [26]. Although many fuzzy decision making problems have been successfully solved by using Per-C [22], [23], [24], [25], [26], [28], its application to FMEA is new, and it provides a useful solution to address the limitations of FMEA. In this paper, the effectiveness of the Per-C-RPN model in FMEA is evaluated using an edible bird nest (EBN) farming task in Sarawak, Borneo Island.

In terms of contributions, a new Per-C-RPN model that preserves intra- and inter- uncertainties of linguistic words in group decision behaviours is introduced. A new FMEA procedure with the proposed Per-C-RPN model is devised. Besides that, the Per-C-RPN model produces RPN scores that can be expressed in both numerals and linguistic words. This advantage provides a better insight pertaining to the risk of a failure mode, in a way that humans can understand the underlying risk semantically in a straightforward manner.

The organisation of this paper is as follows. Preliminaries of the proposed Per-C-RPN model are presented in Section 2. The FMEA procedure with the proposed Per-C-RPN model is explained in Section 3. Simulations with benchmark information [10] are presented in Section 4. A real case study of EBN farming is reported in Section 5. Conclusions are presented in Section 6.

Section snippets

Preliminaries

In this section, the notations and definitions used are presented in Section 2.1, while the background of Per-C and FMEA are described in Sections 2.2 and 2.3, respectively. The proposed Per-C RPN model is presented in Section 2.4.

The FMEA procedure incorporating the proposed Per-C-RPN model

In this section, the FMEA procedure incorporating the proposed Per-C-RPN model is presented, as summarized in Steps (1)–(19), in Fig. 4.

Each step is explained in detail, as follows. To ease the explanation, the EBN case study is used as an example.

Step 1. Develop the scale tables of S, O, and D. In this step, the total number of words used (usually from five to nine [23], [44]) for the scale tables of S, O, and D is determined, respectively. The words used for each S, O, and D are further

Simulation study with a benchmark problem

The background of a benchmark problem [10] is presented in Section 4.1. The simulation results are presented and discussed in Section 4.2. A comparative study between the proposed model and other existing methods is presented in Section 4.3.

Case study on edible bird nest farming

The background of an EBN case study is introduced in Section 5.1. Linguistic grades for assessment are modelled in Section 5.2. Information and data collected are presented in Section 5.3. The outcomes of the proposed Per-C-RPN model are presented in Section 5.4. Finally, a comprehensive discussion is presented in Section 5.5.

Conclusions

A new Per-C-RPN model pertaining to FMEA has been introduced in this study. The FMEA procedure incorporates the proposed Per-C-RPN model. The model adopts linguistic words as the assessment grades, which provide a natural and flexible assessment methodology for FMEA users. Besides that, the inherited uncertainties in the words used for group assessment are preserved in the FMEA procedure. Based on a benchmark problem and an EBN case study, the proposed model has been shown to be useful and

Acknowledgements

The financial support of the FRGS grant (i.e., FRGS/1/2013/ICT02/UNIMAS/02/1), and RACE grants (i.e., RACE/F2/TK/UNIMAS/5) is gratefully acknowledged.

References (51)

  • Y.M. Wang et al.

    Fuzzy TOPSIS method based on alpha level sets with an application to bridge risk assessment

    Expert Syst. Appl.

    (2006)
  • Y.M. Wang et al.

    Risk evaluation in failure mode and effects analysis using fuzzy weighted geometric mean

    Expert Syst. Appl.

    (2009)
  • Z. Zhang et al.

    Risk prioritization in failure mode and effects analysis under uncertainty

    Expert Syst. Appl.

    (2011)
  • K.S. Chin et al.

    Failure mode and effects analysis using a group-based evidential reasoning approach

    Comput. Oper. Res.

    (2009)
  • H. Acosta et al.

    Fuzzy experts on recreational vessels, a risk modelling approach for marine invasions

    Ecol. Modell.

    (2010)
  • K.C. Chai et al.

    A new fuzzy peer assessment methodology for cooperative learning of students

    Appl. Soft Comput.

    (2015)
  • E. Bozdag et al.

    Risk prioritization in failure mode and effects analysis using interval type-2 fuzzy sets

    Expert Syst. Appl.

    (2015)
  • S. Greenfield et al.

    Defuzzification of the discretised generalised type-2 fuzzy set: experimental evaluation

    Inf. Sci.

    (2013)
  • S. Greenfield et al.

    Accuracy and complexity evaluation of defuzzification strategies for the discretised interval type-2 fuzzy set

    Int. J. Approx. Reason.

    (2013)
  • S. Greenfield et al.

    The sampling method of defuzzification for type-2 fuzzy sets: experimental evaluation

    Inf. Sci.

    (2012)
  • W.L. Chang et al.

    Clustering and visualization of failure modes using an evolving tree

    Expert Syst. Appl.

    (2015)
  • K.S. Chin et al.

    Failure mode and effects analysis by data envelopment analysis

    Decis. Support Syst.

    (2009)
  • A.C. Kutlu et al.

    Fuzzy failure modes and effects analysis by using fuzzy TOPSIS-based fuzzy AHP

    Expert Syst. Appl.

    (2012)
  • H.C. Liu et al.

    Failure mode and effects analysis using fuzzy evidential reasoning approach and grey theory

    Expert Syst. Appl.

    (2011)
  • T.L. Saaty et al.

    Why the magic number seven plus or minus two

    Math. Comput. Modell.

    (2003)
  • Cited by (58)

    • From diversity to consensus: Impacts of opinion evolution and psychological behaviours in failure mode and effect analysis

      2022, Applied Soft Computing
      Citation Excerpt :

      As discussed in Section 1, the weights of risk factors are not determined in the traditional FMEA methods. There are many object weights and subject weights methods have been reported, such as the analytic hierarchy process [3,5], and direct assessment method [14]. Gradually, a combination method of subjective and objective weights of risk factors is considered in FMEA.

    View all citing articles on Scopus
    View full text