Projection and bidirectional projection measures of single-valued neutrosophic sets and their decision-making method for mechanical design schemes

Abstract Projection measure is one of important tools for handling decision-making problems. First, the paper proposes projection and bidirectional projection measures between single-valued neutrosophic sets, and then the comparison of numerical examples shows that the bidirectional projection measure is superior to the general projection measure in measuring closeness degree between two vectors. Next, we develop their decision-making method for selecting mechanical design schemes under a single-valued neutrosophic environment. Through the projection measure or bidirectional projection measure between each alternative and the ideal alternative with single-valued neutrosophic information, all the alternatives can be ranked and the best one can be selected as well. Finally, the proposed decision-making method is applied to the selection of design schemes of punching machine and its effectiveness and advantages are demonstrated by comparison with relative methods.


Introduction
Projection measure is a suitable tool for dealing with decision-making problems because it can consider not only the distance but also the included angle between objects evaluated (Xu, 2005;Xu & Da, 2004;Yue, 2012). Therefore, some researchers have successfully applied projection measures to decision-making. For example, Xu and Hu (2010) presented the projection model-based approaches for multiple attribute decision-making problems with intuitionistic and interval-valued intuitionistic fuzzy information. Xu and Cai (2012) proposed projection model-based approaches for intuitionistic fuzzy multiple attribute decision-making problems. Yue (2013) and Zeng, Balezentis, Chen, and Luo (2013) developed projection methods for multiple attribute group decision-making problems with intuitionistic and interval-valued intuitionistic fuzzy information. Yue and Jia (2015) put forward a projection measure for handling a group decision-making problem with hybrid intuitionistic fuzzy information.
As the generalisation of an intuitionistic fuzzy set (IFS) (Atanassov, 1986) and an interval-valued intuitionistic fuzzy (IVIFS) (Atanassov & Gargov, 1989), single-valued neutrosophic sets (SVNSs) (Wang, Smarandache, Zhang, & Sunderraman, 2010) and interval neutrosophic sets (INSs) (Wang, Smarandache, Zhang, & Sunderraman, 2005) are the subclasses of the neutrosophic sets introduced by Smarandache (1998) and are very suitable for describing and handling indeterminate and inconsistent information, which IFSs and IVIFSs cannot describe and deal with, in science and engineering areas. Recently, many researchers have applied SVNSs and INSs to decision-making problems. Some methods have been developed to solve the multiple attribute decision-making problems with SVNS and INS information. For example, the correlation coefficients of SVNSs were used for multiple attribute decision-making (Ye, 2013). A TOPSIS method was extended to interval neutrosophic multiple attribute decision-making problems to rank alternatives (Chi & Liu, 2013). Various similarity measures of SVNSs and INSs were presented and applied to multicriteria (group) decision-making (Ye, 2014a, Ye, 2014b, Ye, 2014c. Singlevalued and interval neutrosophic cross-entropy measures were developed for multiple attribute decision-making problems (Tian, Zhang, Wang, Wang, & Chen, 2016;Ye, 2014d). Some neutrosophic number aggregation operators were proposed and applied to multiple attribute decision-making problems (Liu, Chu, Li, & Chen, 2014;Liu & Wang, 2014). Outranking approaches were applied to multicriteria decision-making problems with simplified neutrosophic sets (including SVNSs and INSs) (Peng, Wang, Zhang, & Chen, 2014;. Then, a multicriteria group decision-making method was introduced under a simplified neutrosophic environment (Peng, Wang, Wang, Zhang, & Chen, 2016). A multiple attribute decision-making method was proposed based on the possibility degree ranking method and ordered weighted aggregation operators of interval neutrosophic numbers (Ye, 2015). Zhang, Ji, Wang, and Chen (2015) proposed an improved weighted correlation coefficient based on integrated weight for INSs and applied it to multicriteria decision-making problems with interval neutrosophic information. Zavadskas, Baušys, and Lazauskas (2015) introduced the sustainable assessment of alternative sites for the construction of a waste incineration plant by the weighted aggregated sum product assessment method with SVNSs. Bausys, Zavadskas, and Kaklauskas (2015) presented a multicriteria decision-making method with SVNSs based on the complex proportional assessment method. A ranking method of single-valued neutrosophic numbers was applied to multiple attribute decision-making problems (Deli & Şubaş, in press). Furthermore, neutrosophic soft sets (Deli, in press), neutrosophic soft multi-sets (Deli, Broumi, & Ali, 2014), and power aggregation operators of multi-valued neutrosophic sets (Peng, Wang, Wu, Wang, & Chen, 2015) were applied to decision-making problems.
However, the existing projection methods cannot deal with decision-making problems with interval neutrosophic information and single-valued neutrosophic information. Furthermore, SVNSs and INSs are scarcely applied in mechanical engineering fields (Ye, in press). Therefore, it is essential to do the research on neutrosophic projection measures and their decision-making method of the mechanical design schemes. Therefore, this paper firstly presents a general projection between SVNSs as a generalisation of the projection measure of IFSs, and then proposes a bidirectional projection measure as an improvement of the general projection measure of SVNSs to overcome the drawback of the general projection in some case. Furthermore, their decision-making method is developed for selecting problems of mechanical design schemes (alternatives) under a single-valued neutrosophic environment.
The rest of the paper is organised as follows. Section 2 briefly describes some basic concepts of neutrosophic sets and SVNSs. Section 3 proposes general projection and bidirectional projection measures between SVNSs and gives their comparison of numerical examples. In Section 4, we develop the general projection and bidirectional projection measures-based decision-making method for selecting mechanical design schemes under a single-valued neutrosophic environment. In Section 5, the proposed decision-making method is applied to the selection of the design schemes of punching machine and its effectiveness and advantages are demonstrated by comparison with relative methods. Finally, Section 6 contains conclusions and future work.

Basic concepts of neutrosophic sets and SVNSs
The neutrosophic set proposed by Smarandache (1998) is a part of neutrosophy and extends the concept of fuzzy sets, interval valued fuzzy set, IFS, and IVIFS from a philosophical point of view. Smarandache (1998) originally gave the definition of a neutrosophic set.
Definition 1. (Smarandache, 1998). Let X be a space of points (objects), with a generic element in X denoted by x. A neutrosophic set N in X is characterised by a truth-membership function T N (x), an indeterminacy-membership function I N (x), and a falsity-membership function F N (x). The functions T N (x), I N (x) and F N (x) are real standard or nonstandard subsets of ] − 0, 1 + [, such that T N (x): X → ] − 0, 1 + [, I N (x): X → ] − 0, 1 + [, and F N (x): X → ] − 0, 1 + [. Hence, the sum of T N (x), I N (x) and F N (x) is no restriction and − 0 ≤ sup T N (x) However, it is difficult to directly apply the neutrosophic set in real science and engineering fields (Wang et al., 2010) due to the nonstandard interval ] − 0, 1 + [. Hence, Wang et al. (2010) introduced a SVNS in the real standard interval [0, 1] as a subclass of a neutrosophic set to suit its engineering applications under an indeterminate and inconsistent environment and gave the definition of a SVNS.
Definition 2. (Wang et al., 2010). Let X be a space of points (objects) with generic elements in X denoted by x. A SVNS N in X is characterised by a truth-membership function T N (x), an indeterminacy-membership function I N (x), and a falsity-membership function F N (x). Then, a SVNS N can be expressed denoted by e = (T, I, F) for short, which is called a single-valued neutrosophic value (SVNV).

Projection and bidirectional projection measures of SVNSs
This section proposes a general projection measure and a bidirectional projection measure for SVNSs.
The projection measure Proj N 2 (N 1 ) can include both the distance and the included angle between N 1 and N 2 . In general, the larger the value of Proj N 2 (N 1 ) is, the closer N 1 is to N 2 .
Based on the extension of the above projection measure of SVNSs, we further propose a bidirectional projection measure between SVNSs below. Definition 5. Let N 1 = {e 11 , e 12 , … , e 1n } and N 2 = {e 21 , e 22 , … , e 2n } be two SVNSs, where e 1j = (T 1j , I 1j , F 1j ) and e 2j = (T 2j , I 2j , F 2j ) (j = 1, 2, … , n) are the j-th SVNVs of N 1 and N 2 respectively. Then is called the bidirectional projection between N 1 and ) are the modules of N 1 and N 2 , respectively, and is the inner product between N 1 and N 2 . The bidirectional projection measure can include not only both the distance and the included angle between N 1 and N 2 but also the bidirectional projection magnitudes between N 1 and N 2 . Obviously, the closer the value of BProj(N 1 , N 2 ) is to 1, the closer the SVNS N 1 is to N 2 . The bidirectional projection measure is a normalised measure, i.e., 0 ≤ BProj(N 1 , N 2 ) ≤ 1.
For the comparison between the general projection measure and the bidirectional projection measure, we consider an example below to show their measuring performance.
Example 1. Let us consider the following two cases: .7 = .8367. In this case, since Proj N 2 (N 1 ) is larger than Proj N 2 (N 3 ), N 1 is much closer to N 2 than N 3 . In fact, since N 3 = N 2 , N 3 should be much closer to N 2 than N 1 , and then Proj N 2 (N 3 ) should be equal to 1. Obviously, the closeness degree between two vectors indicated by the projection measure is not reasonable in this case.
According to Equation (6) N 2 ), N 3 is much closer to N 2 than N 1 and BProj(N 3 , N 2 ) = 1 if and only if N 2 = N 3 . So, the bidirectional projection measure is reasonable and effective.

Decision-making method of mechanical design schemes
In this section, the projection measure and the bidirectional projection measure are used for the multiple attribute decision-making problems of mechanical design schemes with single-valued neutrosophic information.
In the conceptual design stage, designers usually propose various mechanical design schemes (alternatives) according to the function requirements of users and designers. The primal mechanical design schemes can be structured as a set of m alternatives S = {S 1 , S 2 , … , S m }, which must satisfy the requirements of a set of attributes (criteria) R = {R 1 , R 2 , … , R n } by their suitability assessments for fuzzy concept "excellence". The weight w j of the attribute R j (j = 1, 2, … , n) is entered by the decision-maker with w j ∈ [0, 1] and ∑ n j=1 w j = 1. In this case, the characteristic of the alternative S i (i = 1, 2, … , m) with respect to each attribute R j (j = 1, 2, … , n) is expressed by a SVNS form: 1, 2, … , n and i = 1, 2, … , m. For convenience, a basic element in a SVNS S i is denoted by a SVNV e ij = (T ij , I ij , F ij ) for short. Here, the SVNV is usually obtained from the suitability evaluation to which an alternative S i satisfies or does not satisfy an attribute R j by means of a score law or appropriate membership functions in practical applications. Therefore, we can establish an single-valued neutrosophic decision matrix D = (e ij ) m×n .
In multiple attribute decision-making environments, the concept of an ideal alternative has been used to help identify the best alternative in the decision set (Ye, 2014c). Hence, we define the ideal alternative (ideal solution) denoted by the following SVNS: where an ideal SVNV is determined by e ) for j = 1, 2, …, n. Then, by applying Equation (11) or Equation (12) the weighted projection measure or weighted bidirectional projection measure between an alternative S i and the ideal alternative S * is given by The projection measure or the bidirectional projection measure provides the global evaluation for each alternative regarding all attributes. The bigger the measure value of Projw S* (S i ) or Bproj w (S i , S * ) (i = 1, 2, … , m), the better the alternative S i . According to the measure values between the ideal alternative and alternatives, all alternatives can be ranked and the best alternative can be easily selected as well.

Decision-making example of the design schemes of punching machine
This section provides a decision-making example about the selection of the design schemes (alternatives) of punching machine to demonstrate the application and effectiveness of the proposed decision-making method.
In the conceptual design stage, the designers usually need to give a group of primal design schemes to select a better one corresponding to some suitability evaluation for all the primal design schemes. The punching machine generally consists of the reducing mechanism, punching mechanism and feed intermittent mechanism to structure its motion scheme. Therefore, according to its motion scheme, designers propose a set of four potential design schemes (alternatives) S = {S 1 , S 2 , S 3 , S 4 } by their knowledge and experiences, which are shown in Table 1. The chief designer (decision-maker) must take a decision according to the five attributes (criteria): (1) R 1 is the manufacturing cost; (2) R 2 is the structure complexity; (3) R 3 is the transmission effectiveness; (4) R 4 is the reliability; (5) R 5 is the maintainability. The weight vector of the five attributes is w = (.25, .2, .25, .15, .15) T . The four possible alternatives of S i (i = 1, 2, 3, 4) are to be evaluated by the chief designer under the five attributes according to fuzzy concept "excellence" (suitability evaluation), and then the evaluation values are represented by the form of SVNVs.
To indicate the evaluation of an alternative S i (i = 1, 2, 3, 4) with respect to an attribute R j (j = 1, 2, 3, 4, 5), it can be obtained from the questionnaire or score law of a domain expert. For example, when we ask the opinion of the chief designer about an alternative S 1 with respect to an attribute R 1 , he/she may say that the possibility in which the statement is suitable is .75 and the statement is unsuitable is .4 and the degree in which he/she is not sure is .1. By the neutrosophic notation, it can be expressed as e 11 = (.75, .1, .4). Thus, when the four possible alternatives with respect to the above five attributes are evaluated by the chief designer, Thus, the single-valued neutrosophic decision matrix D = (e ij ) 4×5 can be obtained as follows: Then, we utilise the developed approach to obtain the most desirable alternative(s).
For convenient comparison of the example, we introduce the vector similarity measures of SVNSs in (Ye, 2014c) for the decision-making example to show the effectiveness of the proposed projection measures.
Firstly, the proposed projection measures for the decision-making problem are replaced by the cosine measure of Equation (10)  Six bar punching mechanism crank-slider mechanism Dial feed intermittent mechanism Sheave mechanism ratchet feed mechanism Then, the proposed projection measures for the decision-making problem are replaced by another cosine measure, the Dice and Jaccard measures introduced by Ye (2014c): Using Equations (15)-(18), we calculate the vector measures between an alternative S i (i = 1, 2, 3, 4) and the ideal alternative S * , and then all results are also shown in Table 2 for comparative convenience.
In Table 2, obviously the ranking orders based on the projection and bidirectional projection measures are identical, the ranking orders based on the cosine, Dice, and Jaccard measures are identical, and then the ranking orders between the projection and bidirectional projection measures and the cosine, Dice, and Jaccard measures only indicate the difference between S 1 and S 3 . However, for all these measures, S 4 is their optimal choice among all alternatives (mechanical design schemes). In fact, from intuitional viewpoint, the alternative S 4 should also satisfy practical requirements from the designers' experience. Therefore, the proposed projection methods are effective.
Generally, the cosine measures defined in vector space are also not always reasonable in some cases. For example, when e 1 = (T 1 , I 1 , F 1 ) and e 2 = (2T 1 , 2I 1 , 2F 1 ) (e 1 ≠ e 2 ), the values of the cosine measures between e 1 and e 2 are equal to 1. Then, if e 1 = e 2 = (T 1 , I 1 , F 1 ), the cosine measure values of e 1 and e 2 are also equal to 1. Clearly, the cosine measures of Equations (15) and (16) are not reasonable with respect to the decision-making or pattern recognition problems in this case.
However, the bidirectional projection method is superior to the general projection method and the cosine, Dice, Jaccard measures (Ye, 2014c). The reason is the bidirectional projection method for decision-making reveals the following main advantages: (1) The bidirectional projection method is more reasonable than the general projection method because the former can overcome the shortcoming of the latter, then the bidirectional projection measure value is bounded within [0, 1], which is a normalised measure. (2) The bidirectional projection method is more comprehensive than the general projection, cosine, Dice, Jaccard measures because the bidirectional projection can consider not only the distance and the included angle between objects evaluated but also the bidirectional projection magnitudes. (3) The bidirectional projection-based decision-making method provides an effective way for the decision-making of mechanical design schemes under a single-valued neutrosophic environment.

Conclusion
This paper proposed projection and bidirectional projection measures between two SVNSs. Then, the analysis of a numerical example demonstrated that the bidirectional projection measure is superior to the general projection measure in measuring closeness degree between two vectors. Further, a (16) C w (S i , S * ) = n ∑ j=1 w j T ij T * j + I ij I * j + F ij F * j √ T ij 2 + I ij 2 + F ij decision-making method based on the weighted projection or bidirectional project measure was developed and applied to the decision-making problem of mechanical design schemes (alternatives) under a single-valued neutrosophic environment. Through the projection measure or bidirectional projection measure between the ideal alternative and each alternative, we can determine the ranking order of all alternatives and the best one. Finally, a decision-making example on choosing mechanical design schemes of punching machine was provided to demonstrate the applications and effectiveness of the developed approach. Similarly, these proposed projection measures can be also extended to INSs. In the future, we shall further apply the projection and bidirectional projection measures of SVNSs and INSs to group decision-making, medical diagnosis and fault diagnosis problems.

Disclosure statement
No potential conflict of interest was reported by the author.