Advertisement

SN Applied Sciences

, 2:111 | Cite as

Mathematics of uncertainty: an exploration on semi-elliptic fuzzy variable and its properties

  • Palash DuttaEmail author
Research Article
  • 185 Downloads
Part of the following topical collections:
  1. Engineering: Industrial Informatics: Data Analytics in Remote Sensing and Cyber-Physical Systems

Abstract

Uncertainty is more or less encountered in industrial and medical systems as well. Uncertainty theory is an upgraded theory which comprises of possibility measure, necessity measure and credibility measure plays significant role in modelling uncertainty. In connection with uncertainty modeling, a special and intricate fuzzy variable viz., semi-elliptic fuzzy variable (SEFV) is thrashed out here. Subsequently, an attempt has been made to derive possibility, necessity and credibility measure of the SEFV first. Later, some other properties such as expected value, variance, rational upper bound etc are presented and based on that mean and variance ranking of SEFVs are proposed. Afterwards reliability analysis and medical diagnosis cases are carried out which exhibit the efficiency and novelty of the derived SEFV. In this work done, it is observed that the present work has the capability to resolve problems under uncertain complex situations.

Keywords

Uncertainty Fuzzy variable Semi-elliptic fuzzy variable Possibility measure Necessity measure Credibility distribution 

1 Introduction

Uncertainty occurs due to lack of precision, deficiency in data, diminutive sample sizes, foreseeable man-made/artificial mistakes etc., is an unavoidable component of real world problems. To deal with this type of uncertainty fuzzy set theory (FST) [1] is explored. In power system planning reliability investigation is an extremely significant feature. The electrical energy production and consumption are the essential operating characteristic of the power system those are operated simultaneously and consequently, the investigation of reliability obligation for power system is incredibly elevated. Generally probabilistic approaches of reliability investigation are explored. However, due to association of uncertainty in the system classical probability approaches are seemed to be inappropriate and subsequently, fuzzy reliability investigation models are taken into consideration [2]. Some recent applications in reliability investigations can be found in [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]. On the other hand, in medical diagnosis, usually a disease is characterized by several unswervingly perceptible symptoms which persuade the patient to visit a consultant or practitioner. A set of clinical inspections are commenced to make out the incidence of a disease. In the sphere of medical diagnosis, plenty of variables persuade the decision making process and subsequently, discriminate the judgments of the consultant or practitioner. Furthermore, mainly the medical diagnosis quandary engages dealing with uncertainties and so needed to integrate all the information into investigation. Therefore, fuzzy sets are explored to represent uncertainty and to perform medical diagnosis as well [16]. Some recent development in medical diagnosis can also be encountered in [17, 18, 19, 20, 21, 22, 23, 24]. Afterwards Zadeh [25] himself developed possibility theory which was thought to be better to treat uncertainty and further studied by acolyte researchers such as Dubois and Prade [26], Klir [27], Yager [28] etc. Furthermore, Dubois and Prade [29] studied mean value of fuzzy numbers, Ban [30] discussed fuzzy valued measure and conditional expectations of fuzzy numbers, Heilpern [31]studied expected value of fuzzy numbers, Carlsson and Fuller [32] developed possiblistic mean and variance of fuzzy numbers, Chen and Tan [33] further developed mean value and variance of multiplication of fuzzy numbers.

Nevertheless in the absence of self duality measure the earlier studies lead to the exaggeration of the reality. keeping this in mind, Liu and Liu [34] initiated a concept termed as credibility theory. Li and Liu [35] presented a sufficient and necessary condition for credibility measures. Further Liu and Liu [36] systematically studied and developed credibility theory. After that some extended studies on credibility theory can be observed in Liu [37], Zhou et al. [38], Yi et al. [39], Garai et al. [40].

Although various types of fuzzy variables are encountered, however, an exceptional and intricate fuzzy variable SEFV in terms of credibility theory is not deliberated yet. This paper presents an approach to derive possibility, necessity and credibility measure of the SEFV. Furthermore, expected value, variance, rational upper bound etc of SEFV are presented. Then ranking of SEFNs through expected value and variance is proposed. Finally, novelty and applicability has been exhibited by performing reliability analysis and medical diagnosis cases.

2 Preliminaries

Uncertainty is an important as well as unavoidable ingredient of decision making process. Depending on the nature and accessibility of data, information, uncertainty is generally modelled using fuzzy set, possibility theory and Credibility theory.

Definition

Let \(\Theta\) be a non-empty set, and P the power set of \(\Theta\), and Pos is possibility measure. Then, the triplet \((\Theta , P, Pos)\) is known as a possibility space. A fuzzy variable is a mapping from possibility space \((\Theta , P, Pos)\) to the set of real numbers [37, 39].

Definition

Let \(\zeta\) be a fuzzy variable defined on the credibility space \((\Theta , P, Pos)\). Then its membership function (MF) defined from the credibility measure is given by [37, 39]
$$\begin{aligned} \mu _\zeta (x)=min(2Cr{\zeta =1});x\in {\mathbb {R}}. \end{aligned}$$

Definition

The \(\alpha\)-cut of a fuzzy variable A is defined as
$$\begin{aligned} ^{\alpha }A=\left\{ x \in X:\mu _A(x)\ge \alpha \right\} . \end{aligned}$$

Definition

Let A be a fuzzy variable, \(\mu\) be the MF of A, and r be any real number. Then, the possibility measure of A is defined as [25]
$$\begin{aligned} Pos\left\{ A\le r\right\} =\underset{x\le r}{sup}\,\mu _{A}(x) \end{aligned}$$

Definition

Let A be a fuzzy variable, \(\mu\) be the MF of A, and r be any real number. Then, the possibility measure of A is defined as [25]
$$\begin{aligned} Nec\left\{ A\le r\right\} =1-\underset{x>r}{sup}\,\mu _{A}(x) \end{aligned}$$

Definition

A credibility measure (Cr) is a non–negative set function holds the following [35]
  • \(Cr(\Theta )=1\)

  • \(Cr(A)\le Cr(B)\) for whenever \(A\subset B\)

  • \(Cr(A) + Cr(A^c)=1\) for any A

  • \(Cr\left\{ \cup A\right\} =\underset{i}{Sup}Cr \{A_i\}\) for any events \({A_i}\) with \(\underset{i}{Sup}Cr\{A_i\}<0.5\)

If the fuzzy variable A is given by its MF \(\mu\), then
$$\begin{aligned} Cr(A\le r)=\dfrac{1}{2}\left\{ Sup_{x\le r}\mu _A (x) +1 - Sup_{x> r}\mu _A (x) \right\} ,x,y\in {\mathbb {R}} \end{aligned}$$

Definition

The credibility distribution \(\Phi _A:{\mathbb {R}}\rightarrow [0,1]\) of a fuzzy variable A is defined as [36]
$$\begin{aligned} \Phi _A(x)=Cr \{\theta \in \Theta :\zeta (\theta )\le x\}. \end{aligned}$$
That is, the credibility that the fuzzy variable \(\zeta\) takes a value less than or equal to x.

Definition

The credibility density function of credibility distribution defined as \(\phi _A:{\mathbb {R}} \rightarrow [o,\infty )\) of any fuzzy variable A, is a function such that [37]
$$\begin{aligned} \phi _A (x)= \int _{\infty }^x \phi (y)dy, \forall x\in {\mathbb {R}} \end{aligned}$$

Definition

[38] A credibility distribution \(\Phi _A\) of a fuzzy variable A is called regular if it is a continuous and strictly increasing function w.r.t x such that\(0<\Phi _A<1\) and if \(lim_{x\rightarrow -\infty }\Phi _A=0,\) and \(lim_{x\rightarrow \infty }\Phi _A=1.\)

Definition

[38] Let A be a fuzzy variable with a regular credibility distribution \(\Phi _A\), then the inverse function \(\Phi _A^{-1}\) is called the inverse credibility distribution of A.

3 Construction of semi-elliptic fuzzy variable

Let’s consider the general equation of the horizontal ellipse centred at (ab) is
$$\begin{aligned} \dfrac{(x-a)^2}{h^2}+\dfrac{(y-b)^2}{k^2}=1 \end{aligned}$$
To construct a normal semi-circular fuzzy variable (SEFV), it is needed to consider \(b=0\) and \(k=1\). i.e.,
$$\begin{aligned} \dfrac{(x-a)^2}{h^2}+y^2=1 \end{aligned}$$
Thus, the required membership function of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} \mu _A(x)=\sqrt{1-\dfrac{(x-a)^2}{h^2}}, a-h\le x\le a+h \end{aligned}$$
where a indicates the mean/core of the fuzzy variable while h determines width of the fuzzy variable A.
The \(\alpha\)-cut of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} ^\alpha A = \left[ a-h\sqrt{1-\alpha ^2},a+h\sqrt{1-\alpha ^2}\right] \end{aligned}$$
Suppose \(A=S_E(10,7)\) is SEFV representing a uncertain variable. The MF of the SEFV A is
$$\begin{aligned} \mu _A(x)=\sqrt{1-\dfrac{(x-10)^2}{49}}, 3\le x\le 17 \end{aligned}$$
The graphical representation of A is depicted in Fig. 1.
Fig. 1

The SEFV \(A=S_E(10,7)\)

4 Possibility, necessity and credibility measures of SEFV

In this section,the possibility measure, necessity measure and credibility measure of the SEFV are derived.

4.1 Possibility measures of SEFV

Suppose \(A=S_E(a,h)\) is a SEFV.

Then, the possibility measure of (\(A\le x\)) and (\(A\ge x\)) are respectively
$$\begin{aligned} Pos(A \le x)= & {} \left\{ \begin{array}{ll} \,\,\,1, &{}\,\,{\text{if}} \,\,x \ge a, \\ \sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, x<a, \\ \end{array} \right. \\ Pos(A \ge x)= & {} \left\{ \begin{array}{ll} \,\,\,1, &{}\,\,{\text{ if }} \,\, x \le a, \\ \sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, x>a, \\ \end{array} \right. \end{aligned}$$
The possibility measure of the SEFV \(A=S_E(10,7)\) for (\(A\le x\)) and (\(A\ge x\)) are depicted in Figs. 2 and 3 repectively.
Fig. 2

The possibility measure of SEFV \(A=S_E(10,7)\) for (\(A\le x\))

Fig. 3

The possibility measure of SEFV \(A=S_E(10,7)\) for (\(A\ge x\))

4.2 Necessity measures of SEFV

Suppose \(A=S_E(a,h)\) is a SEFV.

Then, the necessity measure of \(A\le x\) and \(A\ge x\) are respectively
$$\begin{aligned} Nec(A \le x)= & {} \left\{ \begin{array}{ll} \,\,\,0, &{}\,\,{\text{ if }} \,\,x \le a, \\ \sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, x>a, \\ \end{array} \right. \\ Nec(A \ge x)= & {} \left\{ \begin{array}{ll} \,\,\,0, &{}\,\,{\text{ if }} \,\, x \ge a, \\ \sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, x<a, \\ \end{array} \right. \end{aligned}$$
The necessity measure of the SEFV \(A=S_E(10,7)\) for (\(A\le x)\) and (\(A\ge x)\) are depicted in Figs. 4 and 5 repectively.
Fig. 4

The necessity measure of SEFV \(A=S_E(10,7)\) for (\(A\le x)\)

Fig. 5

The necessity measure of SEFV \(A=S_E(10,7)\) for (\(A\ge x\))

4.3 Credibility measures of SEFV

Suppose \(A=S_E(a,h)\) is a SEFV.

Then, the credibility measure of \(A\le x\) and \(A\ge x\) are respectively
$$\begin{aligned} Cr(A \le x)= & {} \left\{ \begin{array}{ll} \,\,\,\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\,a-h\le x \le a, \\ 1-\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, a\le x \le a+h, \\ \end{array} \right. \\ Cr(A \ge x)= & {} \left\{ \begin{array}{ll} 1-\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, a-h\le x \le a, \\ \,\,\,\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, a\le x \le a+h, \\ \end{array} \right. \end{aligned}$$
The necessity measure of the SEFV \(A=S_E(10,7)\) for (\(A\le x\)) and (\(A\ge x\)) are depicted in Figs. 6 and 7 repectively.
Fig. 6

The credibility measure of SEFV \(A=S_E(10,7)\) for (\(A\le x\))

Fig. 7

The credibility measure of SEFV \(A=S_E(10,7)\) for (\(A\ge x\))

4.4 Credibility distribution of SEFV

As the credibility distribution \(\Phi _A:{\mathbb {R}}\rightarrow [0,1]\) of a fuzzy variable A is defined as \(\Phi _A(x)=Cr \{\theta \in \Theta :\zeta (\theta )\le x\}\).

Hence, the credibility distribution of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} \Phi _A(x)=\left\{ \begin{array}{ll} \,\,\,\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\,a-h\le x \le a, \\ 1-\dfrac{1}{2}\sqrt{1-\dfrac{(x-a)^2}{h^2}}, &{}\,\,{\text{ if }} \,\, a\le x \le a+h, \\ \end{array} \right. \end{aligned}$$
The credibility distribution function of the SEFV \(A=S_E(10,7)\) is
$$\begin{aligned} \Phi _A(x)=\left\{ \begin{array}{ll} \,\,\,\dfrac{1}{2}\sqrt{1-\dfrac{(x-10)^2}{49}}, &{}\,\,{\text{ if }} \,\,3\le x \le 10, \\ 1-\dfrac{1}{2}\sqrt{1-\dfrac{(x-10)^2}{49}}, &{}\,\,{\text{ if }} \,\, 10\le x \le 17, \\ \end{array} \right. \end{aligned}$$

4.5 Inverse credibility distribution (ICD) of SEFV

The credibility distribution of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} \Phi ^{-1}(\alpha )==\left\{ \begin{array}{ll} a-h \sqrt{1-4\alpha ^2}, &{}\,\,{\text{ if }} \,\,0\le \alpha \le 0.5, \\ a+h \sqrt{1-4(1-\alpha )^2}, &{}\,\,{\text{ if }} \,\, \,\,0.5\le \alpha \le 1, \\ \end{array} \right. \end{aligned}$$

5 Expected value

Using the idea of credibility distribution, Liu and Liu [34] provided the expected value of fuzzy variables. Zhou et al. [38] presented expected value of fuzzy variables via ICD.

5.1 Expected value via credibility distribution

If A is a fuzzy variable then the expected value of A in terms of credibility distribution is defined as [34]
$$\begin{aligned}&E(A)=\int _0^{\infty }Cr\{A\ge r\}dr-\int _{\infty }^0 Cr\{A\le r\}dr\\ \end{aligned}$$
Then, the expected value of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} E(A)&=\int _{0}^{a-h}{dr}+\int _{a-h}^{a}\dfrac{1}{2}\sqrt{1-\dfrac{(r-a)^2}{h^2}}dr\\&\quad + \, \int _{a}^{a+h}\left\{ 1- \dfrac{1}{2}\sqrt{1-\dfrac{(r-a)^2}{h^2}}\right\} dr\\&=(a-h)+h +\int _{a-h}^{a}\dfrac{1}{2}\sqrt{1-\dfrac{(r-a)^2}{h^2}}dr\\&\quad - \, \int _{a}^{a+h} \dfrac{1}{2}\sqrt{1-\dfrac{(r-a)^2}{h^2}}dr\\&=a+\dfrac{\pi h}{8} -\dfrac{\pi h}{8} \\&=a \end{aligned}$$

5.2 Expected value via ICD

If A is a SEFV then the expected value of A in terms of ICD is defined as [38]
$$\begin{aligned} E[A]=\int _0^1\Phi ^{-1}(\alpha )d\alpha \end{aligned}$$
Then, the expected value of the SEFV \(A=S_E(a,h)\) is
$$\begin{aligned} E(A)&=\int _0^{0.5}\left\{ a-h\sqrt{1-4\alpha ^2}\right\} d\alpha \\&\quad +\, \int _{0.5}^{1}\left\{ a+h\sqrt{1-4(1-\alpha )^2}\right\} d\alpha \\&=a -\frac{h}{2}\int _{0}^{1}\sqrt{1-t^2}dt- \frac{h}{2}\int _{1}^{0}\sqrt{1-t^2}dt\\&=a \end{aligned}$$
Thus, in both the approaches, it is obtained that the expected value of the SEFV A is \(e=a\).

Remark

If \(A=S_E(h_1,h_2)\) is a asymmetric SEFV where \(h_1\) and \(h_2\) are left spread and right spread of A. Then, the expected value of A is
$$\begin{aligned} E[A]=a+\frac{\pi }{8}(h_2-h_1). \end{aligned}$$

6 Variance

In this section, variance of SEFV is calculated in terms of regular credibility distribution.

If A is a fuzzy variable with expected value e, then the variance of A is defined as [36]
$$\begin{aligned} V[A]=E[(A-e)^2] \end{aligned}$$
It should be noted that if the expected value e of the fuzzy variable is finite, then the variance satisfies
$$\begin{aligned} V[A]=E[(A-e)^2]=\int _0^{+\infty }Cr\left\{ (A-e)^2\ge r\right\} dr \end{aligned}$$

6.1 Variance of a SEFV

To evaluate variance V[A] of a SEFV A, it is needed to calculate the MF of \((A-e)^2\) first and to find MF of SEFV \(A=S_E(a,h)\), \(\alpha -\)cut technique is explored here.It is already obtained that the expected valued of SEFV is \(e=a\)

The \(\alpha\)-cut of the SEFV \(A=S_E(a,h)\) is \(^\alpha A = [a-h\sqrt{1-\alpha ^2},a+h\sqrt{1-\alpha ^2}]\).

The procedure is presented below.
$$\begin{aligned}&(^{\alpha }A-a)^2\\&\quad =\left[ (a-h\sqrt{1-\alpha ^2}-a)^2,(a+h\sqrt{1-\alpha ^2}-a)^2\right] \\&\quad =\left[ (-h\sqrt{1-\alpha ^2})^2,(h\sqrt{1-\alpha ^2})^2\right] \\&\quad =\left[ h^2(1-\alpha ^2),(h^2(1-\alpha ^2\right] \end{aligned}$$
Now, taking \(x=h^2(1-\alpha ^2)\) gives \(\alpha =\sqrt{1-\dfrac{x}{h^2}}\), \(0 \le x\le h^2\).
Thus, the MF of \((A-e)^2\) is
$$\begin{aligned} \mu _{(A-e)^2} (x)=\sqrt{1-\dfrac{x}{h^2}},0 \le x\le h^2. \end{aligned}$$
Since \(Cr\{(A-e)^2<r\}= \dfrac{1}{2}\{\underset{x<r}{Sup}\mu _{(A-e)^2}(x)+1 -\underset{x\ge r}{sup}\mu _{(A-e)^2}(x)\}\)
$$\begin{aligned} \therefore\, Cr\{(A-e)^2<r\}= \dfrac{1}{2}\left\{ 1+1-\sqrt{1-\dfrac{x}{h^2}}\right\} ,0 \le x\le h^2. \end{aligned}$$
Again, \(Cr\{(A-e)^2\ge r\}=1-Cr\{(A-e)^2<r\}\). Hence, \(Cr\{(A-e)^2\ge r\}={\left\{ \begin{array}{ll}\\ 1-Cr\{(A-e)^2<r\},\quad 0 \le x\le h^2\\ 0, \quad r>h^2 \end{array}\right. }.\) Then, the variance of SEFV \(A=S_E(a,h)\) is
$$\begin{aligned}&V[A]\\&\quad =\int _0^\infty {Cr\{(A-e)^2\ge r\}}dr\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,.\\&\quad =\int _0^{h^2}\dfrac{1}{2}\sqrt{1-\dfrac{r}{h^2}}dr\\&\quad =\dfrac{h^2}{3} \end{aligned}$$
For example, the variance of the SEFV \(A=S_E(10,7)\) is \(\dfrac{7^2}{3}=16.33\).

Remark

If the width h of a SEFV is unit then SEFV will represent a semi-circular fuzzy variable (SCFV). Then, it can be derived that for all SCFV the expected value is 0.33.

7 Rational upper bound of the variance

Yi et al. [39] derived the concept of rational upper bounded of the variance (RUBV) along with some definitions and results in terms of credibility distribution and ICD.

Definition

Let A be a fuzzy variable with credibility distribution \(\Phi\) and finite expected value e. The RUBV is defined as [39]
$$\begin{aligned} \overline{V}[A]=\int _0^\infty (1-\Phi (e+\sqrt{x})\Phi (e-\sqrt{x})) \end{aligned}$$

Definition

Let A be a fuzzy variable with credibility distribution \(\Phi\). If the expected value is e, then RUBV is evaluated as [39]
$$\begin{aligned} \overline{V}[A]=\int _0^1(\Phi ^{-1}(\alpha )-e)^2d\alpha \end{aligned}$$

Corollary

The RUBV of SEFV depends on the width h and it is \(\dfrac{2h^2}{3}.\)

Consider the SEFV \(A=S_E(a,h)\).
$$\begin{aligned}&\int _0^{0.5}h^2(1-4\alpha ^2)d\alpha +\int _{0.5}^1 h^2(1-4(1-\alpha )^2)d\alpha \\&\quad = h^2 + h^2\int _0^{0.5}4\alpha ^2d\alpha -h^2int_{0.5}^14(1-\alpha )^2d\alpha \\&\quad = h^2-\dfrac{h^2}{2}\int _0^1t^2dt+ \dfrac{h^2}{2}\int _1^0t^2dt\\&\quad = h^2-h^2\int _0^1t^2dt\\&\quad = h^2-\dfrac{h^2}{3}\\&\quad =\dfrac{2h^2}{3} \end{aligned}$$

Corollary

Let \(A=S_E(a,h)\) be SEFV. Then, \(\overline{V}[A]=2V[A]\).

Remark

The SCFV also satisfies the above corollary and it is also observed that for all SCFV, the RUBV is 0.66.

Corollary

Suppose \(A=S_E(a_1,h_1)\) and \(B=S_E(a_2,h_2)\) are two SRFV.

Then, \(\overline{V}[A+B]\le 2(\overline{V}[A]+\overline{V}[B]\).

For the two SEFV A and B, the ICV of \(A+B\) is
$$\begin{aligned}&\Phi ^{-1}(\alpha )\\&\quad =\Phi _A^{-1}(\alpha )+\Phi _B^{-1}(\alpha )\\&\quad ={\left\{ \begin{array}{ll} (a_1+b_1)-(h_1+h_2)\sqrt{1-4\alpha ^2},\alpha \le 0.5\\ (a_1+b_1)+(h_1+h_2)\sqrt{1-4(1-\alpha )^2},\alpha >0.5\\ \end{array}\right. } \end{aligned}$$
Now,
$$\begin{aligned}&\overline{V}[A+B]\\&\quad =\int _0^{0.5}[(a_1+b_1)-(h_1+h_2)\sqrt{1-4\alpha ^2}d\alpha \\&\qquad +\, \int _{0.5}^1(a_1+b_1)+(h_1+h_2)\sqrt{1-4(1-\alpha )^2}d\alpha \\&\quad =\dfrac{2(h_1+h_2)^2}{3}. \end{aligned}$$
Again, \(\overline{V}[A]=\dfrac{2(h_1)^2}{3}\) and \(\overline{V}[B]=\dfrac{2(h_2)^2}{3}\).

Since, \(\dfrac{2(h_1+h_2)^2}{3}\le \dfrac{2(h_1)^2}{3}+ \dfrac{2(h_2)^2}{3}.\)

Consequently, \(\overline{V}[A+B]\le 2(\overline{V}[A]+\overline{V}[B]\).

Corollary

Suppose \(A=S_E(a_1,h_1)\) and \(B=S_E(a_2,h_2)\) are two SRFVs.

Then,\(\sqrt{\overline{V}[A+B]}= \sqrt{(\overline{V}[A]}+\sqrt{\overline{V}[B]}\).

Since \(\sqrt{\overline{V}[A+B]}=\sqrt{\dfrac{2(h_1+h_2)^2}{3}}\) and \(\sqrt{\overline{V}[A]}=\sqrt{\dfrac{2h_1^2}{3}}\), \(\sqrt{\overline{V}[B]}=\sqrt{\dfrac{2h_2^2}{3}}\).

Thus, \(\sqrt{\overline{V}[A+B]}= \sqrt{(\overline{V}[A]}+\sqrt{\overline{V}[B]}\).

8 Arithmetic on SEFVs

In this section, basic operations on SEFVs are reviewed and adopted from [41].

Suppose \(A=S_E(a,h)\) and \(B=S_E(b,k)\) are two SEFVs defined on a universe of discourse X.

8.1 Addition

The membership function of \(A+B\) is
$$\begin{aligned} \mu _{(A+B)}&=\sqrt{1-{\Big \{\frac{x-(a+b)}{h_{}+k_{}}\Big \}}^{2}},\\&\quad x\in [(a-h_{})+(b-k_{}), \,(a+h_{})+(b+k_{})] \end{aligned}$$

8.2 Subtraction

The membership function of \(A-B\) is
$$\begin{aligned} \mu _{(A-B)}&=\sqrt{1-{\Big \{\dfrac{x-(a-b)}{h_{}+k_{}}\Big \}}^{2}},\\&\quad x\in [(a-h_{})-(b+k_{}),\,\, (a+h_{})-(b-k_{})] \end{aligned}$$

8.3 Multiplication

The membership function of AB is
$$\begin{aligned} \mu _{AB}(x)&=\sqrt{1-\Big \{\frac{(ak_{}+bh_{})-\sqrt{(ak_{}+bh_{})^{2}-4h_{}k_{}(ab-x)}}{2h_{}k_{}}\Big \}^{}},\\&\quad x \in (a-h_{})(b-k_{})\le x \le (a+h_{})(b+k_{}) \end{aligned}$$

8.4 Division

The membership function of A / B is
$$\begin{aligned} \mu _{\frac{A}{B}}(x)&=\sqrt{1-\Big \{\frac{a-xb}{h_{}+xk_{}}\Big \}^{}},\,\frac{a-h_{}}{b+k_{}}\le x \le \frac{a+h_{}}{b-k_{}} \end{aligned}$$

9 Rank of SEFVs

In this section, ranking of two SEFVs are defined based on the expected value and variance of the SEFVs. Suppose \(A=S_E(a_1,h_1)\) and \(B=S_E(a_2,h_2)\) are two SEFVs.

Then rank of A and B are defined as
  1. I.

    \(A\le B\) if \(E[A]\le E[B]\).

     
  2. II.

    If \(E[A]= E[B]\) then

     

\(A\le B\) if \(V[A]\ge V[B]\)

Example 9.1

Suppose \(A = [-4, 0, 4]\) and \(B = [-2, 0, 2]\) are two fuzzy variables adopted from [42]. It is observed that the approaches [42, 43, 44, 45, 46, 47] fail to compare the fuzzy variables. Reforming these fuzzy variables to SEFVs as \(A = S_E(0,4)\) and \(B = S_E(0,2)\) and applying the present approach it is obtained that \(A\le B\) which is consistent with human intuitions. Here, \(E[A]= E[B]= 0\), but \(V[A]= 5.33 \ge V[B]= 1.33\) and consequently, it can be adopted that \(A\le B\). A details comparison has been presented in Table 1.

Table 1

Ranking of fuzzy variables for example 8.1

Approaches

A

B

Result

Abbasbandy et al. [42]

0

0

\(A\sim B\)

Wang [43]

0

0

\(A\sim B\)

Asady [44]

0.9

0.9

\(A\sim B\)

Asady [45]

0

0

\(A\sim B\)

Asady [46]

0.444

0.444

\(A\sim B\)

Abbasbandy and Hajjari [47]

0.117

0.074

\(A\succ B\)

Present approach

\(E[A]=0.5\)

\(E[B]=0.5\)

 

\(V[A]=5.33\)

V[B]=1.33

\(A\le B\)

Example 9.2

Consider the fuzzy variables \(A=[0.2,0.5,0.8]\) and \(B=[0.4,0.5,0.6]\). The approaches [44, 47, 48, 49, 51] are not applicable to distinguish A and B while [52, 53, 54] and [50] produces illogical output. Here also the general or human intuition is that \(A\le B\). Applying the present approach we obtain \(E[A]= E[B]= 0.5\), but \(V[A]= 0.1 \ge V[B]= 0.033\) which gives \(A\le B\). A detail discussion is presented in Table 2.

Table 2

Ranking of fuzzy variables for example 8.2

Approaches

Decision-level

A

B

Result

Yager [48]

 

0.5

0.5

\(A\sim B\)

Abbhasbandy and Hajjari [47]

 

0.5

0.5

\(A\sim B\)

Asady [44]

 

0.374

0.374

\(A\sim B\)

Chen and Sanguansat [51]

 

0.5

0.5

\(A\sim B\)

Chen et al. [49]

 

0.444

0.444

\(A\sim B\)

Vincent et al. [52, 53]

\(\alpha =1\)

0.117

0.074

\(A\succ B\)

Vincent and Dat [54]

\(\alpha =1\)

0.45

0.35

\(A\succ B\)

Rezvani [50]

 

0.078

0.062

\(A\succ B\)

Present Approach

 

\(E[A]=0.5\)

\(E[B]=0.5\)

 
 

\(V[A]=0.1\)

V[B]=0.033

\(A\le B\)

From the above analysis it can be opined that the present ranking approach has the capability to overcome the drawbacks of the existing approaches.

10 Application of SEFVs

In this segment, application of SEFVs are performed in structural reliability analysis and medical diagnosis. For structural reliability arithmtic of SEFVs are taken into consideration while in medical diagnosis arithmetic as well as ranking of SEFVs are adopted.

10.1 Application in structural reliability

Depictions of the elements of structural analysis are more often SEFVs in some circumstances. In such circumstances structural failure can be evaluated using arithmetic of SEFVs. Consider the following problem of structural failure adopted from Dutta [3].

Example 10.1

Suppose a beam of height \(h=9 mm\), length \(L=1250 mm\) and the force density \(f=78.5\times 10^{-5} kN/mm^3\). The load w, breadth of the beam b and ultimate bending moment \(M_o\) are uncertain input variables represented by SEFVs where \(w=S_E(400,15)kN\), \(b=S_E(40,5)\) and \(M_o=S_E(2.05\times 10^5, 0.05\times 10^5) kN-mm.\) which is depicted in Fig. 8.

The limit state function is \(g(b,f,h,w,M_o,L)=M_o-\left( \dfrac{wL}{4}+\dfrac{fbhL^2}{8}\right)\).

It is needed to evaluate the structural failure of the beam.

Fig. 8

Beam associated with its bending moment

Applying the arithmetic of SEFVs on the problem the value of structural reliability or g is obtained as \(S_E(0.248\times 10^5,0.1658\times 10^5)\).

10.2 Application in medical diagnosis

It is observed that patient’s explanations, medical information even medical assessment process tainted with imprecision/vagueness/uncertainty. On the other hand, knowledge base correlating the symptom-disease relationship encompasses of ambiguity and uncertainty in medical assessment process. Accordingly to deal with such uncertainties FST is being adopted and became most demanding area in medical assessment process. Here, SEFVs are considered to represent uncertain information.

Consider the Patient-symptom and Symptom-disease relations presented in Tables 3 and 4, respectively.
Table 3

Patient-symptom relation

\(R_1\)

\(S_1\)

\(S_2\)

\(S_3\)

\(P_1\)

\(S_E(3,1)\)

\(S_E(6,1)\)

\(S_E(2,1)\)

\(P_2\)

\(S_E(7,1)\)

\(S_E(2,1)\)

\(S_E(4,2)\)

\(P_3\)

\(S_E(2,1)\)

\(S_E(3,1)\)

\(S_E(7,2)\)

Table 4

Symptom-disease relation

\(R_2\)

\(D_1\)

\(D_2\)

\(D_3\)

\(S_1\)

\(S_E(6,2)\)

\(S_E(4,2)\)

\(S_E(3,1)\)

\(S_2\)

\(S_E(2,1)\)

\(S_E(6,1)\)

\(S_E(2,1)\)

\(S_3\)

\(S_E(3,1)\)

\(S_E(2,1)\)

\(S_E(7,2)\)

Table 5

Patient-Disease relation

R

\(D_1\)

\(D_2\)

\(D_3\)

\(P_1\)

\(S_E(36,17,40)\)

\(S_E(52,20,38)\)

\(S_E(35,19,33)\)

\(P_2\)

\(S_E(58,29,35)\)

\(S_E(48,29,28)\)

\(S_E(53,30,28)\)

\(P_3\)

\(S_E(39,23,29)\)

\(S_E(40,23,27)\)

\(S_E(61,32,27)\)

Table 6

Crisp values of the patient-disease relation

R

\(D_1\)

\(D_2\)

\(D_3\)

\(P_1\)

45.0321

59.0686

40.4978

\(P_2\)

60.3562

47.6073

52.2146

\(P_3\)

41.3562

41.5708

59.0365

Using the multiplication and addition of SEFVs the resultant Patient-disease relation is evaluated and presented in Table 5. Then, ranking of SEFVs is adopted to obtain crisp values of resultant SEFVs and presented in Table 6. It should be noted that maximum value in each row indicates that the patient is likely to have the disease. Here, \(\{P_1, P_2, P_3\}\), \(\{S_1, S_2, S_3\}\) and \(\{D_1, D_2,D_3\}\) are the set of patients, symptoms and diseases, respectively. From Table 6, it is clear that the maximum value (the bold value) in 1st row is 59.0686 which associates patient \(P_1\) and disease \(D_2\). That is, patient \(P_1\) is likely to have the disease \(D_2\). Similarly, from 2nd and 3rd row (bold values in the Table 6) it can be opined that patient \(P_2\) is suffering from disease \(D_1\) and patient \(P_3\) is suffering from disease \(D_3\).

11 Conclusions

Uncertainty is an integral part of real world problems such as reliability assessment as well as medical diagnosis problems. To cope with uncertainty a handful number of fuzzy variables are demonstrated yet in literature. However, a special complicated fuzzy variable SEFV is not deliberated well and in this regard here SEFV has been introduced in terms of credibility theory first. Then, some important properties such as possibility measure, necessity measure, credibility measure, credibility distribution and ICD were presented. Afterwards, investigations on expected value of SEFV using credibility distribution and ICD along with variance and RUVB of SEFV have been performed and established relationship between them. Another important concept ranking of SEFVs is introduced based on expected value of SEFV and if it fails then variance of SEFVs concept has been utilized to evaluate order of SEFVs. Comparative numerical illustrations have been presented where results of existing methods and present method were compared and exhibited that present method smoothly over come the limitations of earlier methods. Finally, reliability analysis has been performed using arithmetic of SEFV while a medical diagnosis is performed using arithmetic and rank of SEFVs as well. The present model successfully solves both the problems which exhibits the novelty and applicability of the present model. However, the limitation of this present model is that it can’t work properly when asymmetric SEFVs come into picture. Therefore, as an extension of this work, asymmetric SEFV will be investigated.

Notes

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human or animal subjects.

Supplementary material

42452_2019_1871_MOESM1_ESM.m (0 kb)
Supplementary material 1 (m 0 KB)
42452_2019_1871_MOESM2_ESM.m (0 kb)
Supplementary material 2 (m 0 KB)
42452_2019_1871_MOESM3_ESM.m (0 kb)
Supplementary material 3 (m 0 KB)
42452_2019_1871_MOESM4_ESM.m (0 kb)
Supplementary material 4 (m 0 KB)
42452_2019_1871_MOESM5_ESM.m (0 kb)
Supplementary material 5 (m 0 KB)
42452_2019_1871_MOESM6_ESM.m (0 kb)
Supplementary material 6 (m 0 KB)
42452_2019_1871_MOESM7_ESM.m (0 kb)
Supplementary material 7 (m 0 KB)

References

  1. 1.
    Zadeh LA (1965) Fuzzy sets, Inform. Control 8:338–356MathSciNetzbMATHGoogle Scholar
  2. 2.
    Mohanta DK (2010) Fuzzy reliability evaluations in electric power systems. In: Panigrahi BK, Abraham A, Das S (eds) Computational intelligence in power engineering. Springer, Berlin, pp 103–130Google Scholar
  3. 3.
    Dutta P (2019) Structural reliability analysis with inverse credibility distributions. New Math Nat Comput 15(01):47–63MathSciNetGoogle Scholar
  4. 4.
    Rahimi T, Jahan H, Blaabjerg F, Bahman A, Hosseini S (2019) Fuzzy-logic-based mean time to failure (MTTF) analysis of interleaved DC–DC converters equipped with redundant-switch configuration. Appl Sci 9(1):88Google Scholar
  5. 5.
    Abdolshah M, Samavi A, Khatibi SA, Mamoolraftar M (2019) A review of systems reliability analysis using fuzzy logic. In: Ram M (ed) Advanced fuzzy logic approaches in engineering science. IGI Global, pp 362–377Google Scholar
  6. 6.
    Li H, Nie Z (2018) Structural reliability analysis with fuzzy random variables using error principle. Eng Appl Artif Intell 67:91–99.  https://doi.org/10.1016/j.engappai.2017.08.015 CrossRefGoogle Scholar
  7. 7.
    Gao P, Xie L, Hu W, Liu C, Feng J (2018) Dynamic fuzzy reliability analysis of multistate systems based on universal generating function. Math Probl Eng.  https://doi.org/10.1109/ACCESS.2019.2941508 CrossRefGoogle Scholar
  8. 8.
    Ebrahimnejad A, Jamkhaneh EB (2018) System reliability using generalized intuitionistic fuzzy Rayleigh lifetime distribution. Appl Appl Math 13(1):97–113MathSciNetzbMATHGoogle Scholar
  9. 9.
    Kumar A, Ram M (2018) System reliability analysis based on weibull distribution and hesitant fuzzy set. Int J Math Eng Manag Sci 3(4):513–521Google Scholar
  10. 10.
    Olawoyin R (2017) Risk and reliability evaluation of gas connector systems using fuzzy theory and expert elicitation. Cogent Eng 4(1):1372731Google Scholar
  11. 11.
    Jamali S, Bani MJ (2017) Application of fuzzy assessing for reliability decision making. In: Proceedings of the world congress on engineering and computer science (2017)Google Scholar
  12. 12.
    Lacas E, Santolay JL, Biedermann A (2016) Obtaining sustainable production from the product design analysis. J Clean Prod 139:706–716.  https://doi.org/10.1016/j.jclepro.2016.08.078 CrossRefGoogle Scholar
  13. 13.
    Zhi-gang L, Jun-gang Z, Bo-ying L (2016) Research on reliability evaluation method of complex multistate system based on fuzzy fault tree. Int Conf Fuzzy Theory Appl (Fuzzy).  https://doi.org/10.1109/iFUZZY.2016.8004957 CrossRefGoogle Scholar
  14. 14.
    Rachid B, Hafaifa A, Hadroug N, Boumehraz M (2016) Reliability evaluation based on a fuzzy expert system: centrifugal pump application. Stud Inf Control.  https://doi.org/10.24846/v25i2y201605 CrossRefGoogle Scholar
  15. 15.
    Rizv S, Singh V, Khan A (2016) Fuzzy logic based software reliability quantification framework: early stage perspective (FLSRQF). Procedia Comput Sci 89:359–368.  https://doi.org/10.1016/j.procs.2016.06.083 CrossRefGoogle Scholar
  16. 16.
    Dutta P (2017) Decision making in medical diagnosis via distance measures on interval valued fuzzy sets. Int J Syst Dyn Appl 6(4):63–83Google Scholar
  17. 17.
    Dutta P (2018) Medical diagnosis based on distance measures between picture fuzzy sets. Int J Fuzzy Syst Appl 7(4):15–36Google Scholar
  18. 18.
    Dutta P (2018) Medical diagnosis via distances measures between credibility distributions. Int J Decis Support Syst Technol 10(4):1–16Google Scholar
  19. 19.
    Talukdar P, Dutta P (2018) Disease diagnosis using an advanced distance measure for Intuitionistic Fuzzy Sets. Int Res J Microbiol 7(2):029–042Google Scholar
  20. 20.
    Dutta P, Dash SR (2018) Medical decision making via the arithmetic of generalized triangular fuzzy numbers. Open Cybern Syst.  https://doi.org/10.2174/1874110X01812010001 CrossRefGoogle Scholar
  21. 21.
    Dutta P, Limboo B (2017) Bell-shaped fuzzy soft sets and their application in medical diagnosis. Fuzzy Inf Eng 9(1):67–91Google Scholar
  22. 22.
    Dutta P (2017) Medical diagnosis via distance measures on picture fuzzy sets. AMSE J AMSE IIETA Publ 2017 Ser Adv A 54(2):657–672Google Scholar
  23. 23.
    Dutta P, Doley D (2020) Medical diagnosis under uncertain environment through bipolar-valued fuzzy sets. In: Gupta M, Konar D, Bhattacharyya S, Biswas S (eds) Computer vision and machine intelligence in medical image analysis. Springer, Singapore, pp 127–135Google Scholar
  24. 24.
    Dutta P (2020) A straightforward advanced ranking approach of fuzzy numbers. In: Satapathy SC, Bhateja V, Mohanty JR, Udgata SK (eds) Smart intelligent computing and applications. Springer, Singapore, pp 475–483Google Scholar
  25. 25.
    Zadeh LA (1978) Fuzzy set as a basis for a theory of possibility. Fuzzy Set Syst 1:3–28MathSciNetzbMATHGoogle Scholar
  26. 26.
    Dubois D, Prade H (1988) Possibility theory. Plenum Press, New YorkzbMATHGoogle Scholar
  27. 27.
    Klir JG (1992) On fuzzy set interpretation of possibility theory. Fuzzy Sets Syst 108:263–273MathSciNetzbMATHGoogle Scholar
  28. 28.
    Yager RR (1992) On the specificity of a possibility distribution. Fuzzy Sets Syst 50:279–292MathSciNetzbMATHGoogle Scholar
  29. 29.
    Dubois D, Prade H (1987) The mean value of a fuzzy number. Fuzzy Sets Syst 24:279–300MathSciNetzbMATHGoogle Scholar
  30. 30.
    Ban J (1990) Radon–Nikodým theorem and conditional expectation of fuzzy-valued measures and variables. Fuzzy Sets Syst 34(3):383–392zbMATHGoogle Scholar
  31. 31.
    Heilpern S (1992) The expected value of a fuzzy number. Fuzzy Sets Syst 47:81–86MathSciNetzbMATHGoogle Scholar
  32. 32.
    Carlsson C, Fuller R (2001) On possibilistic mean and variance of fuzzy number. Fuzzy Sets Syst 122:315–326MathSciNetzbMATHGoogle Scholar
  33. 33.
    Chen W, Tan S (2009) On the possibilistic mean value and variance of multiplication of fuzzy numbers. J Comput Appl Math 232(2):327–334MathSciNetzbMATHGoogle Scholar
  34. 34.
    Liu B, Liu YK (2002) Expected value of fuzzy variable and fuzzy expected value model. IEEE Trans Fuzzy Syst 10(4):445–45Google Scholar
  35. 35.
    Li X, Liu B (2006) A sufficient and necessary condition for credibility measures. Int J Uncertain Fuzziness Knowl Based Syst 14(5):527–535MathSciNetzbMATHGoogle Scholar
  36. 36.
    Liu B (2006) A survey of credibility theory. Fuzzy Optim Decis Mak 5(4):387–408MathSciNetzbMATHGoogle Scholar
  37. 37.
    Liu B (2004) Uncertainty theory: a branch of mathematics for modeling human uncertainty. Springer, BerlinGoogle Scholar
  38. 38.
    Zhou J, Yang F, Wang K (2015) Fuzzy arithmetic on LR fuzzy numbers with applications to fuzzy programming. J Intell Fuzzy Syst.  https://doi.org/10.3233/IFS-151712 CrossRefzbMATHGoogle Scholar
  39. 39.
    Yi X, Miao Y, Zhou J, Wang Y (2016) Some novel inequalities for fuzzy variables on the variance and its rational upper bound. J Inequal Appl 1:41MathSciNetzbMATHGoogle Scholar
  40. 40.
    Garai T, Chakraborty D, Roy TK (2017) Expected value of exponential fuzzy number and its application to multi-item deterministic inventory model for deteriorating items. J Uncertain Anal Appl 5(1):8Google Scholar
  41. 41.
    Dutta P, Saikia B (2019) Arithmetic operations on normal semi elliptic intuitionistic fuzzy numbers and their application in decision-making. Granul Comput.  https://doi.org/10.1007/s41066-019-00175-5 CrossRefGoogle Scholar
  42. 42.
    Abbasbandy S, Nuraei R, Ghanbari M (2013) Revision of sign distance method for ranking of fuzzy numbers. Iran J Fuzzy Syst 10(4):101–117MathSciNetzbMATHGoogle Scholar
  43. 43.
    Wang ZX, Liu YJ, Fan ZP, Feng B (2009) Ranking L–R fuzzy number based on deviation degree. Inf Sci 179(13):2070–2077MathSciNetzbMATHGoogle Scholar
  44. 44.
    Asady B (2010) The revised method of ranking LR fuzzy number based on deviation degree. Expert Syst Appl 37(7):5056–5060Google Scholar
  45. 45.
    Asady B, Zendehnam A (2007) Ranking fuzzy numbers by distance minimization. Appl Math Model 31(11):2589–2598zbMATHGoogle Scholar
  46. 46.
    Asady B (2011) Revision of distance minimization method for ranking of fuzzy numbers. Appl Math Model 35(3):1306–1313MathSciNetzbMATHGoogle Scholar
  47. 47.
    Abbasbandy S, Hajjari T (2009) A new approach for ranking of trapezoidal fuzzy numbers. Comput Math Appl 57(3):413–419MathSciNetzbMATHGoogle Scholar
  48. 48.
    Yager RR (1979) Ranking fuzzy subsets over the unit interval. In: 1978 IEEE conference on decision and control including the 17th symposium on adaptive processes, IEEE, pp 1435–1437Google Scholar
  49. 49.
    Chen SM, Munif A, Chen GS, Liu HC, Kuo BC (2012) Fuzzy risk analysis based on ranking generalized fuzzy numbers with different left heights and right heights. Expert Syst Appl 39(7):6320–6334Google Scholar
  50. 50.
    Rezvani S (2015) Ranking generalized exponential trapezoidal fuzzy numbers based on variance. Appl Math Comput 262:191–198MathSciNetzbMATHGoogle Scholar
  51. 51.
    Chen SM, Sanguansat K (2011) Analyzing fuzzy risk based on a new fuzzy ranking method between generalized fuzzy numbers. Expert Syst Appl 38(3):2163–2171Google Scholar
  52. 52.
    Vincent FY, Chi HTX, Shen CW (2013) Ranking fuzzy numbers based on epsilon-deviation degree. Appl Soft Comput 13(8):3621–3627Google Scholar
  53. 53.
    Vincent FY, Chi HTX, Dat LQ, Phuc PNK, Shen CW (2013) Ranking generalized fuzzy numbers in fuzzy decision making based on the left and right transfer coefficients and areas. Appl Math Model 37(16–17):8106–8117MathSciNetzbMATHGoogle Scholar
  54. 54.
    Vincent FY, Dat LQ (2014) An improved ranking method for fuzzy numbers with integral values. Appl Soft Comput 14:603–608Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.Department of MathematicsDibrugarh UniversityDibrugarhIndia

Personalised recommendations