Next Article in Journal
Measuring Information Coupling between the Solar Wind and the Magnetosphere–Ionosphere System
Next Article in Special Issue
Segmentation Method for Ship-Radiated Noise Using the Generalized Likelihood Ratio Test on an Ordinal Pattern Distribution
Previous Article in Journal
Information Entropy-Based Intention Prediction of Aerial Targets under Uncertain and Incomplete Information
Previous Article in Special Issue
Pearle’s Hidden-Variable Model Revisited
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Deng Entropy Weighted Risk Priority Number Model for Failure Mode and Effects Analysis

School of Big Data and Software Engineering, Chongqing University, Chongqing 401331, China
*
Author to whom correspondence should be addressed.
Submission received: 13 February 2020 / Revised: 24 February 2020 / Accepted: 26 February 2020 / Published: 28 February 2020
(This article belongs to the Special Issue Entropy: The Scientific Tool of the 21st Century)

Abstract

:
Failure mode and effects analysis (FMEA), as a commonly used risk management method, has been extensively applied to the engineering domain. A vital parameter in FMEA is the risk priority number (RPN), which is the product of occurrence (O), severity (S), and detection (D) of a failure mode. To deal with the uncertainty in the assessments given by domain experts, a novel Deng entropy weighted risk priority number (DEWRPN) for FMEA is proposed in the framework of Dempster–Shafer evidence theory (DST). DEWRPN takes into consideration the relative importance in both risk factors and FMEA experts. The uncertain degree of objective assessments coming from experts are measured by the Deng entropy. An expert’s weight is comprised of the three risk factors’ weights obtained independently from expert’s assessments. In DEWRPN, the strategy of assigning weight for each expert is flexible and compatible to the real decision-making situation. The entropy-based relative weight symbolizes the relative importance. In detail, the higher the uncertain degree of a risk factor from an expert is, the lower the weight of the corresponding risk factor will be and vice versa. We utilize Deng entropy to construct the exponential weight of each risk factor as well as an expert’s relative importance on an FMEA item in a state-of-the-art way. A case study is adopted to verify the practicability and effectiveness of the proposed model.

1. Introduction

Risk represents the possibility of unintended faults occurring in a system. Uncertainty is a key issue in risk analysis and management [1,2]. Since it was introduced by NASA in 1960s [3], failure mode and effects analysis (FMEA), as a typical bottom-up technique to model and manage potential risks [4], has been extensively applied to various industries, such as medical domain [5,6,7], aircraft landing system [8], assisted reproduction technology [9], automotive industry [10] and so on [11]. In these piratical applications, FMEA is mainly used to ensure that potential risks have been taken into account and dealt with properly during the assessment process. Its most visible result is the documentation of the collective knowledge of cross-functional teams [12]. Traditionally, in this method, risk priority number (RPN) is generally adapted to evaluate and rank the potential failure modes. RPN is a product of three risk factors in FMEA: the probability of occurrence of a failure mode (O), the severity of a failure effect (S), and the probability of a failure being detected (D), which already have been involved in risk assessments. The various components of the target system are prioritized based on RPN scores [13].
Traditional FMEA processes can be summarized as follows, (1) determine the scope of FMEA and assemble a team, (2) identify potential failure failure modes and effects, (4) calculate the RPN of each failure mode and (5) prioritize the failure modes and report the analysis results [3]. Among the processes of the FMEA approach, due to the increasing complexity of system or process, there exists uncertainty when FMEA team members provide their judgments on the identified failure modes. Furthermore, the conventional RPN model is sometimes not that efficient in practical applications [14,15,16]. Flaws of classical RPN can be concluded from three aspects [17,18]. First and foremost, different values of O, S, and D multiplied may be the same RPN value, although they have completely different meanings of risk. For instance, an expert evaluates two different failure modes as (8, 1, 3) and (1, 8, 3) for (O, S, and D), respectively. Though their RPN values are exactly the same, the severity and occurrence of failure 1 is obviously different from that of failure 2. Second, the conventional RPN model ignores the relative importance among O, S, and D. Nevertheless, when applying FMEA this may not be the case with the same weight for the three risk factors. Last but not least, according to the suggested ratings in [19], the RPN score is between 1 and 1000 because each risk factor is on a scale of 1 to 10. However, only 120 of the 1000 numbers can be generated for the product of three risk factors.
Some improved methods have been presented to resolve the aforementioned problems. The uncertainty in complex systems, part of which comes from domain experts’ subjective opinions, needs to be modeled by a reasonable method. In addition to FMEA processes, uncertain information can be processed in practical applications such as distribution networks [20], Mechatronics Engineering [21], nuclear industry [22], infomation fusion and optimization [23], and so on [24]. To deal with the fuzziness in information processing, many theories and methods have been presented, such as Dempster–Shafer evidence theory (DST) [25,26], probability theory [27,28], information entropy [29], fuzzy sets theory [30], and so on [31,32]. For example, in [33], a new risk measure combined with information entropy to measure risk in clustering is proposed. In [34], a measure for modelability, the degree to which an uncertain or fuzzy parameter can be modeled, is presented. Similarly, gray relational projection [35], TOPSIS (technique for order preference by similarity to an ideal solution) [36] and TODIM (an acronym in Portuguese of interactive and multicriteria decision making) [15,37] are utilized to deal with the uncertainty and fuzziness. Because of its capability in uncertain information fusion [38,39,40], DST is used to manage uncertain information in this paper.
Deng entropy, first proposed in [41], aims to measure the uncertain degree of basic probability assignment (BPA) in the framework of DST. Compared with the measure of aggregate uncertainty (AU) [42,43], the ambiguity (AM) [44], and other existing methods for modeling uncertainty, Deng entropy has some advantages [41]. First, Deng entropy satisfies some of the axiomatic requirements that have been further built in order to develop a justifiable measure. The five requirements include range, probabilistic consistency, set consistency, additivity, and subadditivity [45]. Second, Deng entropy is the generalization of Shannon entropy in DST; it provides a promising way to measure uncertain degree and to handle more uncertain information. Third, it is proved that Deng entropy increases monotonously with the increase of the size in a proposition, which also shows it can change more sensitively and rationally than other uncertainty measures. In addition, Deng entropy handles some problems in some other measures in the DST framework [41]. Thus, in this paper, Deng entropy is adopted to measure the uncertainty of experts’ judgments in the framework of DST.
From the perspective of overcoming the shortcomings in traditional RPN, the mean value of RPN (MVRPN) [46], the generalized evidential RPN (GERPN) [47], ambiguity measure weighted risk priority number (AMWRPN) [48], and some other methods [49,50,51] are proposed. For example, in [48], AMWRPN model for FMEA is presented. It processes the weights of risk factors utilizing the fuzziness of experts’ evaluations. Furthermore, fuzzy set theory is an efficient approach to model uncertainty assessments [52,53,54,55]. In [56], the defuzzification of triangular fuzzy numbers is used to determine RPN. In [54], a novel FMEA model utilizing linguistic terms and components’ weights is proposed. In [57], a 2-tuple linguistic variables structure and gray relational analysis are combined to process various assessments of FMEA team members. In addition, in [58], cloud model theory and TOPSIS method are utilized to deal with the risk factor assessments given as linguistic values. However, the previous methods do not take the relative importance of both risk factors and different experts into account. Hence, to resolve this problem, a novel Deng entropy weighted RPN (DEWRPN) model for FMEA is proposed.
In this method, Deng entropy is firstly used to calculate the weights among three risk factors according to the assessments for every failure mode. Then, calculate the weight of each expert by using the corresponding three risk factors’ uncertainty measure-based weights. Finally, calculate the novel RPN based on a novel formula integrating two aforementioned weights. It should be noted that, in comparison with other improved FMEA approaches, there are three desirable properties of the proposed method. First, Deng entropy, as a generalization of Shannon entropy, is an efficient method to obtain the uncertain degree of BPA. Second, the proposed DEWRPN model considers the relative importance of both risk factors and experts, which is more compatible to the real decision-making situation. Last but not least, the value difference of the decision value obtained in this paper is larger than that in [54,59], making the failure mode easier to identify, and the method in this paper is more applicable in complex systems with high requirements for accuracy, which also contributes to decreasing the duplicate value of RPN.
The rest of this paper is organized as follows. The preliminaries are introduced in Section 2. In Section 3, a new Deng entropy weighted risk priority number approach for FMEA model, named DEWRPN, is proposed. Then, the DEWRPN-based FMEA approach is used to analyze a case study of steel production process in a steel factory in Section 4. The conclusion of this paper is shown in Section 5.

2. Preliminaries

In this section, some basic concepts in the DST [25,26], FMEA [3], and Deng entropy [41] are introduced.

2.1. Dempster–Shafer Evidence Theory

DST originated in the work of Dempster using probabilities with upper and lower bounds [25] and Shafer established the basic probability assignment function (BPA) on the framework of discernment [26]. DST develops on the foundation of the Bayesian theory of probabilities. It can represent and process uncertain information effectively. Reasoning and decision-making can be carried out with incomplete or conflicting pieces of evidence even if there is a lack of prior information. Formally, the definitions in DST are provided as follows.
Definition 1.
Let Ω be a set of mutually exclusive and collectively exhaustive elements H i , indicated by
Ω = H 1 , H 2 , , H i , , H N .
The power set of Ω composed with 2 N propositions is called the Frame of Discernment (FOD), denoted as 2 Ω :
2 Ω = , H 1 , H 2 , , H N , H 1 , H 2 , , H 1 , H 2 , , H i , , Ω ,
where each element is a proposition andis an empty set. In addition, each single set that contains only one element in FOD is called singleton.
Definition 2.
A basic probability assignment (BPA) (also called mass function) is a mapping for elements in 2 Ω to the interval [0,1], formally defined by:
m : 2 Ω 0 , 1 ,
which satisfies the following conditions:
m = 0 , A Ω m A = 1 ,
where “A” symbolizes any subset of Ω, which is A Ω . If A , the BPA function m ( A ) represents how strongly the evidence supports the hypothesis A. If m ( A ) > 0 , the A in the frame of discernment is called a focal element and the set of all the focal elements is named a body of evidence (BOE).
Definition 3.
A BPA m can also be represented by the belief function B e l or the plausibility function P l , defined as follows:
B e l A = B A m B    ,    P l A = B A m B .
Definition 4.
Two pieces of evidence in the frame of discernment Ω indicated as m 1 and m 2 . A focal elements of m 1 is described as B and that of m 2 is presented as C. The Dempster’s combination rule can be defined as follows:
m 1 , 2 ( A ) = m 1 ( B ) m 2 ( C ) = B , C Ω , B C = A m 1 ( B ) × m 2 ( C ) 1 B C = , B C = A m 1 ( B ) × m 2 ( C ) ,
where a coefficient K is defined as follows:
K = B C = m 1 ( B ) × m 2 ( C ) .
Sometimes, the K is defined as a conflict coefficient between two BOEs.
Definition 5.
Let m be a BPA defined on the frame of discernment X. The pignistic probability distribution B e t P m , called as such by Smets and Kennes [60], is defined for all the subsets of X by
B e t P m A = B X m B | A B | | B | ,
where | A | is the cardinality of set A.

2.2. FMEA

FMEA is an analytical tool in reliability and risk management, which has been proved to be remarkably effective and applied extensively in many fields such as risk evaluation [58], decision making [61,62,63], and so on. A series of failure modes with high priorities are determined in conventional FMEA [64]. Facilitating the identification of potential failures in the design or process of products or systems, the process of applying an FMEA can be briefly concluded in Figure 1, where the FMEA process is summarized in detail as eleven steps [65].
  • Step 1. Understand the operating properly process of a system.
  • Step 2. Subdivide the system as subsystems and/or assemblies according to system’s characteristics.
  • Step 3. Use various analysis tools to identify system’s components and how they relate to each other.
  • Step 4. List all components for each subsystem and/or assembly.
  • Step 5. Identify the key point that may have an impact on the performance of components and cause unexpected failures.
  • Step 6. Determine the failure modes in detail for evaluation.
  • Step 7. Domain experts evaluate the probability of failures qualitatively.
  • Step 8. Process the uncertainty in experts’ evaluations and calculate the RPN.
  • Step 9. Prioritize failure modes identified with different risk levels.
  • Step 10. Pay more attention to failure modes with high priorities and develop an appropriate action plan mainly from two aspects:
    -
    Preventive actions for avoiding a failure.
    -
    Compensatory actions for minimizing losses brought by an unwanted failure.
  • Step 11. Make a summary of the previous analysis with tabular form.
Traditionally, during the application process of FMEA, prioritizing failure modes based on RPN is a key step. A failure mode with a higher RPN value is concerned with being more critical than that with a lower RPN.
Definition 6.
The RPN consists of three factors: the probability of occurrence of a failure mode ( O ) , the severity of a failure effect ( S ) , and the probability of a failure being detected ( D ) . RPN can be defined as follows:
R P N = O × S × D .
Generally, each risk factor can be measured with 10 ranking levels from 1 to 10. For instance, Table 1 shows the suggested criteria of rating for the occurrence O of a failure in FMEA. Similarly, the severity S of a failure effect and the detectability D of a failure can be mapped to an integer from 1 to 10. More details can be found in [46].

2.3. Deng Entropy

Deng entropy (DE) is proposed by Deng yong, which satisfies five axiomatic requirements: range, probabilistic consistency, set consistency, additivity, and subadditivity. Compared with Hohle’s confusion measure, Yager’s dissonance measure, and Klir and Ramer’s discord, Deng entropy can increase monotonously with the rise of the size of focal sets for mass functions in DST [41].
Definition 7.
DE is defined as follows [41]:
D E m = A X m A log 2 m A 2 | A | 1 ,
where m is the a mass function defined on the frame of discernment X, and A is a focal element of m. | A | stands for the cardinality of A.
As a generalization of Shannon entropy, Deng entropy is similar with it formally. However, the belief of each focal element A is divided by a term, 2 | A | 1 , which represents the potential number of states in A (the empty set is not included). In addition, Deng entropy is actually a type of composite measure through a simple transformation.
Definition 8.
D E m = A X m A log 2 2 | A | 1 m A log 2 m A ,
where the term, A X m A log 2 2 | A | 1 , can be regarded as a measure of a total nonspecificity in the mass function m, and the remainder of Equation (11), A X m A log 2 m A , is the measure of discord of the mass function among various focal elements.

3. Deng Entropy Weighted Risk Priority Number for FMEA

A new RPN model is proposed based on the DE in the DST framework to handle the relative weight of each risk factor and each expert in the FMEA model. The proposed method is composed of the following seven steps concluded from the new DEWRPN model for FMEA process in Figure 2:
  • Step 1. List all failure modes (FMs) and corresponding cause of these failure modes in the system on a basis of historical data, past experiences, and expert opinions.
  • Step 2. FMEA experts give assessments on each FMEA item for three indicators O , S , D . In DST framework, the collected subjective assessments are modeled as BPAs.
  • Step 3. Original BPAs need to be normalized for an optimal transformation. Usually, the pignistic distribution is taken advantage of.
  • Step 4. Measure the uncertainty degree of each risk factor by DE, which is also regarded as the risk factor weight.
Definition 9.
According to the definition of Equation (10), the uncertainty degree of each risk factor for the ith expert can be calculated as follows:
D E O i = A O i X m A log 2 m A 2 | A | 1 , D E S i = A S i X m A log 2 m A 2 | A | 1 , D E D i = A D i X m A log 2 m A 2 | A | 1 ,
where A is a focal element of the mass function of the mathematical expression of evaluation corresponding to the required risk factor. X is the frame of discernment of risk factors, and X = O , S , D . m A is a mass function defined on the frame of discernment X. The integrated values of all the risk factors, O i , S i and D i , by the ith expert can be calculated respectively as follows:
O i = j = 1 3 R j m j O i , S i = j = 1 3 R j m j S i , D i = j = 1 3 R j m j D i ,
where assume j=(1, 2, 3) and R 1 = 1 , R 2 = 2 , R 3 = 3 . R j is the corresponding rating value of evaluation grades determined differently in different applications. Thus, the value range of index j and R j varies from case to case. m j O i , m j S i , and m j D i are the mass functions of the corresponding rating values evaluated by the ith expert.
  • Step 5. Calculate the absolute expert weights based on corresponding uncertainty degrees of three risk factors.
Definition 10.
The absolute weight for ith expert is defined as follows.
ω i = D E O i + D E S i + D E D i
  • Step 6. Calculate the novel RPNs for each FM aggregating both risk factor weights and expert weights.
Definition 11.
Based on the the assessments given by n ( n 1 ) independent experts in an FMEA team, the Deng entropy weighted risk priority number (DEWRPN) for each failure mode is defined as follows:
D E W R P N = i = 1 n ω i i = 1 n ω i O i e D E O i × S i e D E S i × D i e D E D i
where D E · measures the uncertainty degree of the assessment as BPA, which is provided by an expert to the corresponding risk factor, and ω · weighs the absolute experts’ weights as Definition 10. Correspondingly, O i , S i , and D i respectively aggregate the ith expert’s evaluating values for each risk factor: O, S, and D. Moreover, e D E · expresses the relative weight of each risk factor based on the uncertainty degree measured by Equation (12). Meanwhile, ω i i = 1 n ω i is the relative weight of ith expert compared with all experts involved in FMEA.
  • Step 7. Prioritize these FMs by the ranking of novel RPNs.
It should be noted that, in the DEWRPN approach, the weight for each risk factor is obtained objectively and independently from relevant evaluations. It is only connected with the assessments themselves. For example, when two experts evaluate the same failure mode with different values, the weights for three factors are surely different. Besides, the DE-based expert weight is also strongly related to assessments for three risk factors, which suggests that when the same expert assesses two failure modes, if the assessments are different, the expert weights in these two failure modes are different. In other words, the assessments from different experts have no impact on other expert weights. In short, the weight, whether for three risk factors or for experts, is objective and independent. It is measured by the uncertainty degree in assessments. This is the biggest advantage of DEWRPN.

4. Application and Discussion

In this section, an application of DEWRPN model for FMEA is utilized to illustrate its practicability and efficiency. The result and comparisons are discussed as well.

4.1. Application

The practicability and effectiveness of DEWRPN model for FMEA are verified by a case study in [66]. The evaluations for ten failure modes of this case study are given by Deshpande and Modak [66]. In this section, these failure modes are processed and ranked by the proposed method. Furthermore, the priority of an FMEA item is related to their occurrence probability, severity of the related effects, and detection to each failure mode.
The aforementioned seven steps for applying DEWRPN model for FMEA are described as follows.
Step 1. List the FMs by the system versus three indexes as shown in Table 2.
Step 2. From the adopted case study, the assessments given by FMEA experts on each FMEA item for three indicators O , S , D are presented in [67]. In the DST framework, the collected subjective assessments are modeled as BPAs. The data are shown in detail in Table 3. From the table, there are ten FMs F M 1 , F M 1 , , F M 10 , three criteria O , S , D , and three FMEA experts E 1 , E 2 , E 3 . Meanwhile, each judgment is based on three evaluation grades g o o d , m o d e r a t e , p o o r . In this case study, the first expert evaluates the severity of F M 1 with a belief of 80% that F M 1 is not serious, a belief of 10% that it may take a risk of failing, and a belief of 10% that it is serious.
Step 3. After examination, it is worth noting that the sum of the assessed percentages for each FMEA item factor in the original data is equal to 1, which can also be described as α 1 + α 2 + α 3 = 1 . Thus, original BPAs do not need to be normalized for an optimal transformation. The data in Table 3 can be utilized directly.
Step 4. According to the definition of Equation (12), the uncertainty degree of each risk factor for each expert can be measured by DE. The exponential weight based on it is also regarded as the risk factor weight. In addition, when aggregating assessment rating value for each risk factor, the three evaluation grades g o o d , m o d e r a t e , p o o r are assigned with 1, 2, and 3 respectively. Thus, the advantage of Equation (13) can be adopted. The greater score of the original rating value for a factor means greater risk for this factor. To simplify the explanation of the experiment results, the ith FMEA item is also described as F M i . The calculation result for FMEA item F M 1 is shown in Table 4.
Step 5. Calculate the absolute expert weights based on corresponding uncertainty degrees of three risk factors according to Equation (10).
Step 6. Calculate the novel RPNs for each FM aggregating both risk factor weights and expert weights utilizing Equation (15). The calculation results are presented in Table 5. The DEWRPN-based priorities for ten FMs in the sheet steel production process are sorted from high to low, shown as follows: F M 4 F M 3 F M 8 F M 7 F M 1 F M 2 F M 10 F M 5 F M 6 F M 9 , where ≻ means the former has a higher risk priority than the latter.
Step 7. Prioritize these FMs by the ranking of novel RPNs. Using the results from Table 5, analyze the results and assign finite resource to the FMEA item with high priority. In this case study, F M 4 with highest RPN value is the riskiest in ten FMs. According to Table 2, more attention may be paid to the burn-out electrode and if its cooler works properly.

4.2. Discussion

In order to illustrate the validation of this novel DEWRPN model for FMEA, the results obtained for ten FMs using the proposed method are compared with the sorting results in [54,59], where the methods and experiment results are rational. The comparison result is displayed in Figure 3. Analyzing the comparison result, the ranking results yielded by DEWRPN are consistent with that obtained from Behnam Vahdani et al.’s method [54] and from Li and Chen’s method [59]. Note that, among all the 10 FMEA items, F M 4 is considered to be the riskiest one, and the other failure modes have similar risk levels. Therefore, DEWRPN’s result is practical and rational.
However, there exists little difference for F M 7 and F M 10 . For F M 7 and F M 10 , the evaluated probability distribution for the risk factors reflects the uncertainty degree, which also has an impact on risk factor weights and expert weights. Thus, F M 7 and F M 10 assessed by the proposed method and Beham Vahdani’s method have a higher priority than that processed by Li and Chen’s method as indicated in Figure 3. In general, compared with Behnam Vahdani et al. and Li and Chen’s method, DEWRPN model for FMEA can generate lower priorities for F M 1 , F M 5 , F M 6 , F M 9 . The reason for this can be concluded to be that the uncertainty degrees of FMEA experts’ evaluations are captured and depicted as risk factor weights and expert weights by DEWRPN. Furthermore, from the perspective of absolute RPN values, the broken line composed of RPNs obtained by Li and Chen’s method has exactly the same trend with that of the improved RPNs according to Figure 4. On the other hand, this proves the correctness of the proposed method. Moreover, by analyzing these curves, a desirable feature by which the proposed method can amplify the difference between each FM to yield a more stable and distinguishable ranking is shown.
In comparison with the aforementioned methods, DEWRPN emphasizes the corresponding weights of risk factors and absolute expert weights, which are independently obtained from the evaluations for each FMEA item. Intuitively, if the same expert evaluates two FMEA items differently, the expert weights are certainly different. This feature embodies the objectiveness of the proposed method. Considering the relative importance of both risk factors and experts, it overcomes the limitations of the traditional RPN. In addition, higher variances of RPN values of different FMs not only make RPN values more distinguishable for the prioritization process, they also make this method more applicable to some complex systems for higher accuracy in contrast to Li and Chen’s method. Last but not least, in DEWRPN, constructing RPN with risk factor weights and expert weights utilizing subjective assessments is a comprehensive way to generate RPN-based priority for each FMEA item.
In the future work, according to [3], the calculation of multiplication is meaningless on ordinal scales because the three risk factors (O, S, and D) are evaluated based on discrete ordinal scales of measure. So in the following work, we will consider finding an improved way to combine these three risk factors.

5. Conclusions

In the traditional FMEA, the shortcomings in computation of RPN have been criticized. A novel RPN model for the FMEA approach named DEWRPN is proposed in this paper. It constructs new RPN values by utilizing subjective assessments in a more comprehensive way to generate the final ranking values. Intuitively, if different experts evaluate FMEA items differently, the relative weights of experts are different. DEWRPN transforms the uncertain degree of subjective assessments, which is measured by Deng entropy, as the relative importance of risk factors and FMEA experts. A case study on the sheet steel production process verifies the practicability and efficiency of the novel method.
Future research can apply the proposed method to other industry fields. In addition, some recently proposed uncertainty measures in the evidence theory can be taken into consideration to address the different kinds of uncertainties in subjective assessments [68].

Author Contributions

Conceptualization, Y.T.; Methodology, H.Z. and Y.T.; Validation, H.Z.; Writing—original draft, H.Z.; Writing—review and editing, H.Z. and Y.T. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Doshi, H.; Kumar, P.; Yerramilli, V. Uncertainty, capital investment, and risk management. Manag. Sci. 2018, 64, 5769–5786. [Google Scholar] [CrossRef] [Green Version]
  2. Yazdi, M.; Kabir, S. Fuzzy evidence theory and Bayesian networks for process systems risk analysis. Hum. Ecol. Risk Assess. Int. J. 2020, 26, 57–86. [Google Scholar] [CrossRef] [Green Version]
  3. Liu, H.C. FMEA Using Uncertainty Theories and MCDM Methods. In FMEA using Uncertainty Theories and MCDM Methods; Springer Singapore: Singapore, 2016; pp. 13–27. [Google Scholar]
  4. Peeters, J.; Basten, R.J.; Tinga, T. Improving failure analysis efficiency by combining FTA and FMEA in a recursive manner. Reliab. Eng. Syst. Saf. 2018, 172, 36–44. [Google Scholar] [CrossRef] [Green Version]
  5. Batbayar, K.; Takács, M.; Kozlovszky, M. Medical device software risk assessment using FMEA and fuzzy linguistic approach: Case study. In Proceedings of the 2016 IEEE 11th International Symposium on Applied Computational Intelligence and Informatics (SACI), Timisoara, Romania, 12–14 May 2016; pp. 197–202. [Google Scholar]
  6. Shi, S.; Fei, H.; Xu, X. Application of a FMEA method combining interval 2-tuple linguistic variables and grey relational analysis in preoperative medical service process. IFAC-PapersOnLine 2019, 52, 1242–1247. [Google Scholar] [CrossRef]
  7. Gaur, K. Systematic and quantitative assessment and application of FMEA and Lean six sigma for reducing non productive time in operation theatre of a Tertiary Care Hospital in a metropolis. Perioper. Care Oper. Room Manag. 2019, 16, 100075. [Google Scholar] [CrossRef]
  8. Daneshvar, S.; Yazdi, M.; Adesina, K.A. Fuzzy smart failure modes and effects analysis to improve safety performance of system: Case study of an aircraft landing system. Qual. Reliab. Eng. Int. 2020. [Google Scholar] [CrossRef]
  9. Intra, G.; Alteri, A.; Corti, L.; Rabellotti, E.; Papaleo, E.; Restelli, L.; Biondo, S.; Garancini, M.P.; Candiani, M.; Viganò, P. Application of failure mode and effect analysis in an assisted reproduction technology laboratory. Reprod. Biomed. Online 2016, 33, 132–139. [Google Scholar] [CrossRef] [Green Version]
  10. De Aguiar, D.C.; Salomon, V.A.P.; Mello, C.H.P. An ISO 9001 based approach for the implementation of process FMEA in the Brazilian automotive industry. Int. J. Qual. Reliab. Manag. 2015, 32, 589–602. [Google Scholar] [CrossRef]
  11. Sang, A.J.; Tay, K.M.; Lim, C.P.; Nahavandi, S. Application of a Genetic-Fuzzy FMEA to Rainfed Lowland Rice Production in Sarawak: Environmental, Health, and Safety Perspectives. IEEE Access 2018, 6, 74628–74647. [Google Scholar] [CrossRef]
  12. Mzougui, I.; El Felsoufi, Z. Proposition of a modified FMEA to improve reliability of product. Procedia CIRP 2019, 84, 1003–1009. [Google Scholar] [CrossRef]
  13. Panchal, D.; Kumar, D. Risk analysis of compressor house unit in thermal power plant using integrated fuzzy FMEA and GRA approach. Int. J. Ind. Syst. Eng. 2017, 25, 228–250. [Google Scholar]
  14. Balaraju, J.; Raj, M.G.; Murthy, C.S. Fuzzy-FMEA risk evaluation approach for LHD machine-A case study. J. Sustain. Min. 2019, 18, 257–268. [Google Scholar] [CrossRef]
  15. Huang, J.; Li, Z.S.; Liu, H.C. New approach for failure mode and effect analysis using linguistic distribution assessments and TODIM method. Reliab. Eng. Syst. Saf. 2017, 167, 302–309. [Google Scholar] [CrossRef]
  16. Huang, Z.; Jiang, W.; Tang, Y. A new method to evaluate risk in failure mode and effects analysis under fuzzy information. Soft Comput. 2018, 22, 4779–4787. [Google Scholar] [CrossRef]
  17. Wang, Y.M.; Chin, K.S.; Poon, G.K.K.; Yang, J.B. Risk evaluation in failure mode and effects analysis using fuzzy weighted geometric mean. Expert Syst. Appl. 2009, 36, 1195–1207. [Google Scholar] [CrossRef]
  18. Liu, H.C.; Liu, L.; Liu, N. Risk evaluation approaches in failure mode and effects analysis: A literature review. Expert Syst. Appl. 2013, 40, 828–838. [Google Scholar] [CrossRef]
  19. Chang, K.H. Evaluate the orderings of risk for failure problems using a more general RPN methodology. Microelectron. Reliab. 2009, 49, 1586–1596. [Google Scholar] [CrossRef]
  20. Barutcu, I.C.; Karatepe, E.; Boztepe, M. Impact of harmonic limits on PV penetration levels in unbalanced distribution networks considering load and irradiance uncertainty. Int. J. Electr. Power Energy Syst. 2020, 118, 105780. [Google Scholar] [CrossRef]
  21. Yin, X. Adaptive backstepping control for maximizing marine current power generation based on uncertainty and disturbance estimation. Int. J. Electr. Power Energy Syst. 2020, 117, 105329. [Google Scholar] [CrossRef]
  22. Cheng, Y.; Hao, C.; Li, F. Uncertainty quantification of fuel pebble model and its effect on the uncertainty propagation of nuclear data in pebble bed HTR. Ann. Nucl. Energy 2020, 139, 107286. [Google Scholar] [CrossRef]
  23. Karve, P.M.; Guo, Y.; Kapusuzoglu, B.; Mahadevan, S.; Haile, M.A. Digital twin approach for damage-tolerant mission planning under uncertainty. Eng. Fract. Mech. 2020, 225, 106766. [Google Scholar] [CrossRef]
  24. Pecht, M.; Shibutani, T.; Wu, L. A reliability assessment guide for the transition planning to lead-free electronics for companies whose products are RoHS exempted or excluded. Microelectron. Reliab. 2016, 62, 113–123. [Google Scholar] [CrossRef]
  25. Dempster, A.P. Upper and Lower Probabilities Induced by a Multi-valued Mapping. Ann. Math. Stat. 1967, 38, 325–339. [Google Scholar] [CrossRef]
  26. Shafer, G. A Mathematical Theory of Evidence; Princeton University Press: Princeton, NJ, USA, 1976. [Google Scholar]
  27. Dubois, D.; Prade, H. Possibility theory, probability theory and multiple-valued logics: A clarification. Ann. Math. Artif. Intell. 2001, 32, 35–66. [Google Scholar] [CrossRef]
  28. Sabahi, F.; Akbarzadeh-T, M.R. Introducing validity in fuzzy probability for judicial decision-making. Int. J. Approx. Reason. 2014, 55, 1383–1403. [Google Scholar] [CrossRef]
  29. Shannon, C.E. A mathematical theory of communication. ACM SIGMOBILE Mob. Comput. Commun. Rev. 2001, 5, 3–55. [Google Scholar] [CrossRef]
  30. Zadeh, L.A. Fuzzy sets. Inf. Control. 1965, 8, 338–353. [Google Scholar] [CrossRef] [Green Version]
  31. Zhou, X.; Tang, Y. A Note on Incomplete Information Modeling in the Evidence Theory. IEEE Access 2019, 7, 166410–166414. [Google Scholar] [CrossRef]
  32. Li, Z.; Xiao, F.; Fei, L.; Mahadevan, S.; Deng, Y. An evidential failure mode and effects analysis using linguistic terms. Qual. Reliab. Eng. Int. 2017, 33, 993–1010. [Google Scholar] [CrossRef]
  33. Xu, X.h.; Luo, X. Information entropy risk measure applied to large group decision-making method. Soft Comput. 2019, 23, 4987–4997. [Google Scholar] [CrossRef]
  34. Hawer, S.; Schönmann, A.; Reinhart, G. Guideline for the classification and modelling of uncertainty and fuzziness. Procedia CIRP 2018, 67, 52–57. [Google Scholar] [CrossRef]
  35. Baghery, M.; Yousefi, S.; Rezaee, M.J. Risk measurement and prioritization of auto parts manufacturing processes based on process failure analysis, interval data envelopment analysis and grey relational analysis. J. Intell. Manuf. 2018, 29, 1803–1825. [Google Scholar] [CrossRef]
  36. Dutta, B.; Singha, T.; Goh, M.; Lamata, M.T.; Verdegay, J.L. Post factum analysis in TOPSIS based decision making method. Expert Syst. Appl. 2019, 138, 112806. [Google Scholar] [CrossRef]
  37. Pramanik, S.; Mallick, R. TODIM strategy for multi-attribute group decision making in trapezoidal neutrosophic number environment. Complex Intell. Syst. 2019, 5, 379–389. [Google Scholar] [CrossRef] [Green Version]
  38. Liu, P.; Zhang, X. A new hesitant fuzzy linguistic approach for multiple attribute decision making based on Dempster–Shafer evidence theory. Appl. Soft Comput. 2020, 86, 105897. [Google Scholar] [CrossRef]
  39. Zhang, H.; Deng, Y. Engine fault diagnosis based on sensor data fusion considering information quality and evidence theory. Adv. Mech. Eng. 2018, 10, 1687814018809184. [Google Scholar] [CrossRef] [Green Version]
  40. Biswas, B.; Ghosh, S.K.; Bhattacharyya, S.; Platos, J.; Snasel, V.; Chakrabarti, A. Chest X-ray enhancement to interpret pneumonia malformation based on fuzzy soft set and Dempster–Shafer theory of evidence. Appl. Soft Comput. 2020, 86, 105889. [Google Scholar] [CrossRef]
  41. Deng, Y. Deng entropy. Chaos, Solitons & Fractals 2016, 91, 549–553. [Google Scholar]
  42. Harmanec, D.; Klir, G.J. Measuring total uncertainty in Dempster-Shafer theory: A novel approach. Int. J. Gen. Syst. 1994, 22, 405–419. [Google Scholar] [CrossRef]
  43. Maeda, Y.; Nguyen, H.T.; Ichihashi, H. Maximum entropy algorithms for uncertainty measures. Int. J. Uncertainty, Fuzziness -Knowl.-Based Syst. 1993, 1, 69–93. [Google Scholar] [CrossRef]
  44. Jousselme, A.L.; Liu, C.; Grenier, D.; Bossé, É. Measuring ambiguity in the evidence theory. IEEE Trans. Syst. Man, -Cybern.-Part Syst. Humans 2006, 36, 890–903. [Google Scholar] [CrossRef]
  45. Klir, G.J.; Wierman, M.J. Uncertainty-based Information: Elements of Generalized Information Theory; Physica-Verlag: New York, NY, USA, 2013; Volume 15. [Google Scholar]
  46. Yang, J.; Huang, H.Z.; He, L.P.; Zhu, S.P.; Wen, D. Risk evaluation in failure mode and effects analysis of aircraft turbine rotor blades using Dempster–Shafer evidence theory under uncertainty. Eng. Fail. Anal. 2011, 18, 2084–2092. [Google Scholar] [CrossRef]
  47. Zhou, D.; Tang, Y.; Jiang, W. A modified model of failure mode and effects analysis based on generalized evidence theory. Math. Probl. Eng. 2016, 2016. [Google Scholar] [CrossRef]
  48. Tang, Y.; Zhou, D.; Chan, F.T. AMWRPN: Ambiguity measure weighted risk priority number model for failure mode and effects analysis. IEEE Access 2018, 6, 27103–27110. [Google Scholar] [CrossRef]
  49. Liu, H.C.; Wang, L.E.; You, X.Y.; Wu, S.M. Failure mode and effect analysis with extended grey relational analysis method in cloud setting. Total. Qual. Manag. Bus. Excell. 2019, 30, 745–767. [Google Scholar] [CrossRef]
  50. Su, X.; Deng, Y.; Mahadevan, S.; Bao, Q. An improved method for risk evaluation in failure modes and effects analysis of aircraft engine rotor blades. Eng. Fail. Anal. 2012, 26, 164–174. [Google Scholar] [CrossRef]
  51. Hu, Y.P.; You, X.Y.; Wang, L.; Liu, H.C. An integrated approach for failure mode and effect analysis based on uncertain linguistic GRA–TOPSIS method. Soft Comput. 2019, 23, 8801–8814. [Google Scholar] [CrossRef]
  52. Kutlu, A.C.; Ekmekcioglu, M. Fuzzy failure modes and effects analysis by using fuzzy TOPSIS-based fuzzy AHP. EXpert Syst. Appl. 2012, 39, 61–67. [Google Scholar] [CrossRef]
  53. Song, W.; Ming, X.; Wu, Z.; Zhu, B. Failure modes and effects analysis using integrated weight-based fuzzy TOPSIS. Int. J. Comput. Integr. Manuf. 2013, 26, 1172–1186. [Google Scholar] [CrossRef]
  54. Vahdani, B.; Salimi, M.; Charkhchian, M. A new FMEA method by integrating fuzzy belief structure and TOPSIS to improve risk evaluation process. Int. J. Adv. Manuf. Technol. 2015, 77, 357–368. [Google Scholar] [CrossRef]
  55. Seiti, H.; Hafezalkotob, A.; Najafi, S.; Khalaj, M. A risk-based fuzzy evidential framework for FMEA analysis under uncertainty: An interval-valued DS approach. J. Intell. Fuzzy Syst. 2018, 35, 1419–1430. [Google Scholar] [CrossRef]
  56. Iranzadeh, S. Investigating the relationship between RPN parameters in fuzzy PFMEA and OEE in a sugar factory. J. Loss Prev. Process. Ind. 2019, 60, 221–232. [Google Scholar]
  57. Liu, H.C.; Li, P.; You, J.X.; Chen, Y.Z. A novel approach for FMEA: Combination of interval 2-tuple linguistic variables and gray relational analysis. Qual. Reliab. Eng. Int. 2015, 31, 761–772. [Google Scholar] [CrossRef]
  58. Liu, H.C.; Wang, L.E.; Li, Z.; Hu, Y.P. Improving risk evaluation in FMEA with cloud model and hierarchical TOPSIS method. IEEE Trans. Fuzzy Syst. 2018, 27, 84–95. [Google Scholar] [CrossRef]
  59. Li, Z.; Chen, L. A novel evidential FMEA method by integrating fuzzy belief structure and grey relational projection method. Eng. Appl. Artif. Intell. 2019, 77, 136–147. [Google Scholar] [CrossRef]
  60. Smets, P.; Kennes, R. The transferable belief model. Artif. Intell. 1994, 66, 191–234. [Google Scholar] [CrossRef]
  61. Chanamool, N.; Naenna, T. Fuzzy FMEA application to improve decision-making process in an emergency department. Appl. Soft Comput. 2016, 43, 441–453. [Google Scholar] [CrossRef]
  62. Sharma, R.K.; Sharma, P. System failure behavior and maintenance decision making using, RCA, FMEA and FM. J. Qual. Maint. Eng. 2010, 16, 64–88. [Google Scholar] [CrossRef]
  63. Zhang, H.; Dong, Y.; Palomares-Carrascosa, I.; Zhou, H. Failure mode and effect analysis in a linguistic context: A consensus-based multiattribute group decision-making approach. IEEE Trans. Reliab. 2018, 68, 566–582. [Google Scholar] [CrossRef] [Green Version]
  64. Kang, J.; Sun, L.; Sun, H.; Wu, C. Risk assessment of floating offshore wind turbine based on correlation-FMEA. Ocean. Eng. 2017, 129, 382–388. [Google Scholar] [CrossRef]
  65. Pillay, A.; Wang, J. Modified failure mode and effects analysis using approximate reasoning. Reliab. Eng. Syst. Saf. 2003, 79, 69–85. [Google Scholar] [CrossRef]
  66. Deshpande, V.S.; Modak, J. Application of RCM to a medium scale industry. Reliab. Eng. Syst. Saf. 2002, 77, 31–43. [Google Scholar] [CrossRef]
  67. Sankar, N.R.; Prabhu, B.S. Modified approach for prioritization of failures in a system failure mode and effects analysis. Int. J. Qual. Reliab. Manag. 2001, 3, 324–336. [Google Scholar] [CrossRef]
  68. Xiao, F. Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy. Inf. Fusion 2019, 46, 23–32. [Google Scholar] [CrossRef]
Figure 1. The flowchart of failure mode and effects analysis (FMEA).
Figure 1. The flowchart of failure mode and effects analysis (FMEA).
Entropy 22 00280 g001
Figure 2. The flowchart of Deng entropy weighted risk priority number (DEWRPN) model for FMEA.
Figure 2. The flowchart of Deng entropy weighted risk priority number (DEWRPN) model for FMEA.
Entropy 22 00280 g002
Figure 3. The ranking of failure modes based on the proposed method in comparison with existing methods.
Figure 3. The ranking of failure modes based on the proposed method in comparison with existing methods.
Entropy 22 00280 g003
Figure 4. The RPN values generated by the proposed method in comparison with that of [59].
Figure 4. The RPN values generated by the proposed method in comparison with that of [59].
Entropy 22 00280 g004
Table 1. Suggested criteria of rating for occurrence of a failure in FMEA.
Table 1. Suggested criteria of rating for occurrence of a failure in FMEA.
RatingProbability of OccurrencePossible Failure Rate
10Extremely high (almost inevitable)≥0.500
9Very high0.3333
8Repeated failure0.1250
7High0.0500
6Moderately high0.0125
5Moderate0.0025
4Relatively low0.0005
3Low0.0000667
2Remote0.0000067
1Nearly impossible≤0.0000066
Table 2. The FMEA of the sheet steel production process in Guilan steel factory.
Table 2. The FMEA of the sheet steel production process in Guilan steel factory.
No.Failure Mode (FM)Cause of Failure
A 1 Nonacceptable formationNoncondective scrap
A 2 Nipple thread pittedProper coverage not obtained
A 3 Arc formation lossLeakage of water, proper gripping loss
A 4 Burn-out electrodeCooler not working properly
A 5 Breaking of house of pipeWearing of pipe due to use
A 6 Problem in movement of armSever leakage
A 7 Refractory damageDue to slag
A 8 Formation of steamRoof leak
A 9 Refractory line damageBy hot gas
A 10 Movement of roof stopJam of plunger in unloader valve
Table 3. Group belief structure judgment of the sheet steel production process.
Table 3. Group belief structure judgment of the sheet steel production process.
FMsExpertsSeverityOccurrenceDetectability
F M 1 E 1 (0.8, 0.1, 0.1)(0.1, 0.2, 0.7)(0.2, 0.5, 0.3)
E 2 (0.7, 0.0, 0.3)(0.0, 0.4, 0.6)(0.3, 0.4, 0.3)
E 3 (0.8, 0.2, 0.0)(0.1, 0.4, 0.5)(0.2, 0.5, 0.3)
F M 2 E 1 (0.7, 0.1, 0.2)(0.1, 0.2, 0.7)(0.8, 0.1, 0.1)
E 2 (0.7, 0.0, 0.3)(0.0, 0.4, 0.6)(0.7, 0.0, 0.3)
E 3 (0.6, 0.4, 0.0)(0.1, 0.4, 0.5)(0.8, 0.2, 0.0)
F M 3 E 1 (0.8, 0.1, 0.1)(0.0, 0.1, 0.9)(0.2, 0.5, 0.3)
E 2 (0.9, 0.0, 0.0)(0.0, 0.2, 0.8)(0.3, 0.4, 0.3)
E 3 (0.7, 0.3, 0.0)(0.1, 0.0, 0.9)(0.2, 0.5, 0.3)
F M 4 E 1 (0.4, 0.4, 0.2)(0.0, 0.1, 0.9)(0.1, 0.2, 0.7)
E 2 (0.3, 0.5, 0.2)(0.0, 0.2, 0.8)(0.0, 0.4, 0.6)
E 3 (0.4, 0.4, 0.2)(0.1, 0.0, 0.9)(0.1, 0.4, 0.5)
F M 5 E 1 (0.4, 0.4, 0.2)(0.2, 0.4, 0.4)(0.7, 0.0, 0.3)
E 2 (0.3, 0.5, 0.2)(0.2, 0.4, 0.4)(0.8, 0.2, 0.0)
E 3 (0.4, 0.4, 0.2)(0.1, 0.5, 0.4)(0.6, 0.3, 0.1)
F M 6 E 1 (0.4, 0.4, 0.2)(0.2, 0.4, 0.4)(0.7, 0.0, 0.3)
E 2 (0.3, 0.5, 0.2)(0.2, 0.4, 0.4)(0.8, 0.2, 0.0)
E 3 (0.4, 0.4, 0.2)(0.1, 0.5, 0.4)(0.6, 0.3, 0.1)
F M 7 E 1 (0.4, 0.4, 0.2)(0.2, 0.4, 0.4)(0.1, 0.2, 0.7)
E 2 (0.5, 0.5, 0.0)(0.2, 0.4, 0.4)(0.0, 0.4, 0.6)
E 3 (0.6, 0.4, 0.0)(0.1, 0.5, 0.4)(0.1, 0.4, 0.5)
F M 8 E 1 (0.8, 0.1, 0.1)(0.0, 0.1, 0.9)(0.2, 0.5, 0.3)
E 2 (0.9, 0.0, 0.0)(0.0, 0.2, 0.8)(0.3, 0.4, 0.3)
E 3 (0.7, 0.3, 0.0)(0.1, 0.0, 0.9)(0.2, 0.5, 0.3)
F M 9 E 1 (0.4, 0.4, 0.2)(0.2, 0.4, 0.4)(0.7, 0.0, 0.3)
E 2 (0.3, 0.5, 0.2)(0.2, 0.4, 0.4)(0.8, 0.2, 0.0)
E 3 (0.4, 0.4, 0.2)(0.1, 0.5, 0.4)(0.6, 0.3, 0.1)
F M 10 E 1 (0.7, 0.0, 0.3)(0.2, 0.4, 0.4)(0.2, 0.5, 0.3)
E 2 (0.8, 0.2, 0.0)(0.4, 0.0, 0.6)(0.3, 0.4, 0.3)
E 3 (0.6, 0.3, 0.1)(0.4, 0.0, 0.6)(0.2, 0.5, 0.3)
Table 4. DE and aggregated rating values of each expert for F M 1 .
Table 4. DE and aggregated rating values of each expert for F M 1 .
FM 1 Expert 1Expert 2Expert 3
D E · D E O 1 = 1.1568 D E S 1 = 0.9219 D E D 1 = 1.4855 D E O 2 = 0.9710 D E S 2 = 0.8813 D E D 2 = 1.5710 D E O 3 = 1.3610 D E S 3 = 0.7219 D E D 3 = 1.4855
Rating O 1 = 2.6000 S 1 = 1.3000 D 1 = 2.1000 O 2 = 2.6000 S 2 = 1.6000 D 2 = 2.0000 O 3 = 2.4000 S 3 = 1.200 D 3 = 2.1000
Table 5. DEWRPN values and risk priority ranking.
Table 5. DEWRPN values and risk priority ranking.
NO. FM 1 FM 2 FM 3 FM 4 FM 5 FM 6 FM 7 FM 8 FM 9 FM 10
DEWRPN 1.7991 1.7786 2.3715 2.8101 1.5904 1.5904 1.8694 2.3715 1.5726 1.7310
Rank56218943107

Share and Cite

MDPI and ACS Style

Zheng, H.; Tang, Y. Deng Entropy Weighted Risk Priority Number Model for Failure Mode and Effects Analysis. Entropy 2020, 22, 280. https://0-doi-org.brum.beds.ac.uk/10.3390/e22030280

AMA Style

Zheng H, Tang Y. Deng Entropy Weighted Risk Priority Number Model for Failure Mode and Effects Analysis. Entropy. 2020; 22(3):280. https://0-doi-org.brum.beds.ac.uk/10.3390/e22030280

Chicago/Turabian Style

Zheng, Haixia, and Yongchuan Tang. 2020. "Deng Entropy Weighted Risk Priority Number Model for Failure Mode and Effects Analysis" Entropy 22, no. 3: 280. https://0-doi-org.brum.beds.ac.uk/10.3390/e22030280

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop