1- Introduction

2- Related work

3- Problem formulation

4- Experiments and results

5- Conclusions

References

Abstract

Cloud computing refers to providing computing services and resources over the Internet. The cloud provider is an organization or company which offers the services to the consumer at different levels of features and characteristics. However, as the growth of cloud services as well as cloud service providers are increasing rapidly, it is becoming a challenge for consumers to choose the best service provider based on their requirements. In this paper, we propose a method to help the consumer to answer this question. A hybrid multi criteria decision method (MCDM) is developed to evaluate and rank cloud service providers from Smart data. Furthermore, this method considers the interdependencies and relations between the performance measurements. The hybrid method consists of two components: (i) clustering the providers using k-means algorithm to consolidate cloud service providers with similar features and (ii) applying MCDM methods using DEMATEL-ANP to rank clusters and make a final decision. The proposed method also considers the existing workloads of the organization as well as assigns different importance and weights for a set of criteria by clustering the cloud service providers using k-means algorithm. A simulation on the MATLAB was performed to evaluate the proposed method, and the results indicate how the proposed hybrid approach can provide an accurate and efficient way to select the best providers.

Introduction

Cloud computing is emerging technology in computing resources allocation over the Internet. This model of computing provides important benefits to the organizations by relieving them of low-level tasks related to setting up IT infrastructure as well as allowing organizations to begin small resources and increase these resources on demand, thus enabling more time for innovation and the creation of business value. This technology is a model for providing a pool of on demand resources like software as a service SaaS, infrastructure as a service IaaS, and platform as a service PaaS [1]. Nowadays, migrating the applications, data, and/or infrastructure to a cloud is a confirmed challenging process. Certain obstacles prevent cloud computing from offering its full features. Those obstacles are related to the fact that current applications, data, and/or infrastructure have specific needs and configurations that must be realized by the cloud provider [1, 2]. From an organization’s perspective, each system has its own configuration performance and factors that affect workload parameters. Furthermore, each cloud service provider offers its services with various features such as performance, cost, and security. Therefore, it is challenging for organizations to choose the most appropriate cloud service provider in the presence of these multiple features.

ANALYTIC network process (ANP), firstly brought out by Saaty in 1996 [1], allows one to include all the factors and criteria that have impacts on making a best decision. It is of great use in assisting the mind to organize its thoughts and experiences and to elicit judgments recorded in memory and qualify them in the form of priorities. It allows the decision makers (DMs) to represent diverse opinions after discussion and debate. The ANP, as a generalized model of the analytic hierarchy process (AHP) [2], not only can solve the AHP problems, but also can tackle interdependent relationships within multiple criteria decision making (MCDM) problems by replacing hierarchies with networks. It is based on deriving ratio scale measurements and can be used to allocate resources according to their ratio-scale priorities. It provides a way to the input judgments and measurements to derive ratio scale priorities for the distribution of influences among the factors and groups of factors in the decision making problem. The steps of ANP involve the following steps: ① constructing control hierarchy and network and identifying feedbacks or dependences among the elements and clusters, ② constructing the pairwise comparison matrices regarding to the elements and clusters, ③ deriving local priorities and constructing unweighted supermatrix with local priorities, ④ adjusting the unweighted supermatrix to the weighted supermatrix (also called as column stochastic matrix), ⑤ limiting the weighted supermatrix by raising it to an arbitrarily large power and calculating the limit priorities from it , and ⑥ deriving the final priorities of the alternatives. Although the traditional ANP is widely used in management science and operations research, we shall not ignore its drawback: the DMs cannot guarantee that the judgements valued by them are exact and crisp. During the pairwise comparison procedure, the DMs usually are required to give the value of the preference relation by crisp numbers based on the knowledge and experience they owned. However, only by the crisp numbers cannot express the DMs’ uncertainty on the preference relation. If the experts cannot clearly comprehend the problem, they are unwilling to give their judgements by crisp values, and then we cannot successfully solve the problem. In order to overcome this shortcoming of ANP, Mikhailov and Singh [3] made enormous strides in the direction of fuzzy ANP (FANP) and its applications in MCDM. So far, the FANP have been applied in many areas, such as evaluating region agricultural drought risks [4], selecting container ports [5], selecting social media platform [6], choosing supplier [7], evaluating ship maneuverability [8], and determining the importance of hospital information system adoption factors [9].

The objective of this study is to satisfy the market demand for the product systematized design and to help a designer select the optimal product concept and assess related problems. It is expected to reduce the risk of having various products in a unified form while the market demand may possibly be neglected. This study aims at satisfying various customer requirements on the highly competitive market under limited time and resources. The most relevant design conceptual idea can be determined and a designer can effectively master the relationship between consumer needs and critical product series attributes. However, during the selection and assessment of design concepts, the subjective perception typically affects the selection and thus a designer may fail to make a clear and definite decision. With the benefit of a strategic decision and assessment method by using the decision-making trial and evaluating laboratory (DEMATEL) approach, it is different from conventional multi-criteria decision making method which usually assumes that the criteria are independent of each other. Wu and Tsai [1] proposed that the DEMATEL approach can further help decision-makers confirm the causal relationships among designs and also proposed an evaluation method for the measurement and a strategic decision testing by DEMATEL. Tseng and Wu et al. [1,3] proposed to integrate different aspects of thinking into a single reason and influence. For combining with hybrid evaluation method of the strategic decision testing and assessment by the DEMATEL technique, Shen [2] used the ANP method to build related techniques for selecting models. Wu [7] combined the ANP method with the DEMATEL method based on solution packages so as to assist in the necessary assessments and selection of knowledge management strategies. In addition, the DEMATEL method can transform the relationship between a conversion factor and its standardisation effect into a visualized interpretive structural model. It can also serve as a standard method for determining the dependency or correlation between a factor and its effect. Tseng [3] proposed a mixed ANP method for the analysis from the dependency aspect. A strategic decision testing and assessment by the DEMATEL approach was used to clarify the correlation and the fuzzy set theory was used for the assessment of the uncertainty.

The location of military engineering(LME) is an important aspect of battlefield environment construction in the future information condition. A success of LME, directly related to the realization of the national strategy, the rational use of national resources and the effective use of defense funds, operation command comprehensive coordination, the optimized allocation of battlefield target , etc. In the case of the construction of a large integrated strategic homeport, it is indispensable to support a growing fleet and the country’s frontier maritime strategy. In existing port or the port group, to chose the basic conditions are relatively good, and had the potential to extend the facility for key construction and make it to its homeport, not only conform to the requirements of the future naval strategy and berthing ship’s security requirements, but also can save construction investment, the navy can make full use of the facility results on the basis of the existing building, make more power for port shore facilities maintenance, update and supplement. In other words, a suitable location of the military port˄LMP˅ is of great significance to improve the strategic value, to give full play of security efficiency and to improve the military economic benefits. At present, there are a lot of existing location assessment methods, Xiao Ding et al. apply the method of entropy weight and ideal point method to select a site for civil port [1]. Huang Mingsheng and Taih-Cherng Lirn adopt AHP method to decided the weights of site selection [2] [3]; Liu Linhu et al. using principal component analysis to select candidate military logistics base, on the basis of the maximum coverage model are used to determine a location decision [4]. The mufti-objective decision model of military engineering location is constructed by Weng Dongfeng [5]. Halpern proposed a two-level standard plane model for the actual problem of site selection for the first time[6]. Ayfer and Ergin has applied ELECTRE method to the site selection of container ports [7]. It is required that when the above methods in constructing a model of the whole evaluation system has a complete cognitive structure, and the relationship between the accurate model covers the elements of information is less, but through the study found that the construction of port planning and site selection evaluation indicator is broad, the relevance between the indicators are hidden, the elements in the system is not absolutely independent, but mutual influence and interdependence. In order to make the result of LMP more reasonable and credible, this paper proposes a Bayesian-ANP-based assessment method, which is fused expert group of judgment through comprehensive consideration all experts preference indicator correlation of evaluation, and optimize the process of the construction of the ANP model, overcome the subjective arbitrariness of independent evaluation, improve the reliability of the ANP model.

With the advent of competitive markets, companies have acknowledged the crucial necessity of progressing the maintenance system of the companies. Diverse sections of industries face numerous strategic decisions to attain the best feasible and realistic answers in risky and competitive environments. These decisions are extremely significant regarding the technology development, diversity of customers in addition to public pressure. Maintenance is a significant task in any manufacturing system and has a critical role in attaining the organization’s objectives (Seiti, Hafezalkotob and Fattahi, 2017). With the introduction of different maintenance systems, diverse maintenance strategies are applied to attain more added values. In this regard, numerous studies have been conducted on maintenance strategies (Alabdulkarim et al., 2015; Shaaban and Awni, 2014; Graisa and Al-Habaibeh, 2011; Savsar, 2010). Choosing an appropriate maintenance policy is a critical activity in manufacturing systems as it affects the performance of companies’ equipment. This practice is a common multi-criteria decision-making (MCDM) problem. An MCDM problem includes both perceptible and imperceptible factors (Galankashi et al., 2013; Ziaei et al., 2013; Dargi et al., 2014). Although different studies have been conducted on maintenance policy selection, only a few attempts have been made to see the MCDM approaches in conjunction with the effect of cost, risk and added value. In this regard, recognizing, incorporating and selecting the right maintenance policy to efficiently screen and assess the current situation of the company are a challenge for many researchers. Cost, risk and added value are very influential in maintenance planning. However, their simultaneous consideration is less investigated in the previous studies. This is somewhat due to intrinsic difficulties connected with complex models of maintenance policy selection. Therefore, considering these factors in maintenance planning is useful and affects the performance of companies’ equipment. The main justification to use a fuzzy-based method to tackle the problem of this research is the inherent struggle of comparing different policies using decision makers’ comments. The decision makers are comfortable to compare these strategies in fuzzy environment instead of deterministic scales. In addition, Fuzzy Analytic Network Process (FANP) is a suitable method to study the relations between decision elements in hierarchical structures. Therefore, developing an FANP model to rank and select the best maintenance policy for different equipment of an acid manufacturing plant is the main objective of this study. As the main justification to apply this method, FANP considers the uncertainty of data and the relation of criteria and alternatives which are very unique compared to other MCDM approaches.

1. Introduction

Due to urbanization and industrialization, China has been the biggest contributor to emissions since 2007 [1]. Industry, transportation and buildings are the three main fields of energy consumption in China, and consumption in the building sector in China has occupied more than 33% of the national energy consumption [2]. Currently, the total building area in China is near 40 billion m2 , and by 2020, the total building area will reach 70 billion m2 [2]. Meanwhile, more than 90% of existing buildings are high energy consumption, thus making China’s existing building energy consumption per unit area approximately 2 times that of developed countries in same environmental conditions [3]. Building Energy Efficiency Retrofit (BEER) provides excellent opportunities to reduce energy consumption in buildings and encourages the implementations of other sustainability practices, such as environmental protection, rational resource use, and occupants’ healthcare [4]. Building Energy Efficiency Retrofit (BEER) is a process to reduce buildings’ operational energy use by applying certain approaches of the building envelope and mechanical systems improvements while keeping the buildings’ indoor environment and comfort [5]. Thus, China should accelerate the energy efficiency improvements of existing buildings. The Energy Service Companies (ESCOs) industry, a recently emergent industry, provides a market mechanism called energy performance contracting (EPC) to deliver energy efficiency projects. EPC is a financing package that includes energy savings guarantees, associated designs and installation services for energy efficiency retrofit projects [6–9]. EPC was introduced to China in 1996 in partnership with the World Bank and the Global Environment Fund. The program aims to introduce EPC, improve energy efficiency, reduce greenhouse gas emissions, and protect the environment in China. In 1998, three pilot ESCOs were created: the Beijing ESCO, the Liaoning ESCO, and the Shandong ESCO [8]. In 2004, The Energy Management Company Association (EMCA) was created to facilitate the operation of EPC and development of the ESCOs industry in China. From then on, the number of ESCOs experienced a rapid growth period. The numbers of ESCOs, investments and production have increased over all these years (Figs. 1 and 2). There has surely been a significant average annual growth rate of 34.91% in the number of ESCOs from 2004 to 2013. In this millennium, the Chinese government has established a series of policies to support the application of EPC and the development of ESCOs. Obviously, the government’s attention brings huge developmental opportunities for ESCOs. However, such a number cannot represent a healthy energy service market since there are some inevitable obstacles leading to a difficult situation for ESCOs, especially in building sector [10]. Most of the ESCOs in China remain small. According to the EMCA, the financial capacity of Chinese ESCOs is limited: 60% of the ESCOs have less than 10 million RMB of registration capital; 20% have more than 50 million RMB of registration capital; and approximately 50% of ESCOs have less than 100 employees [11]. Moreover, in China, it is difficult for small companies with little capital to obtain loans or other financial support, thereby making it difficult for ESCOs to obtain financing [12–14].

Landslides as a type of mass movements involve slow or fast movement of soil and stone materials, or both on the slopes downwards, under the force of gravity (Crosta and Clague, 2009). Landslides are known as one the most common geological disasters which cause damages and casualties worldwide (Bianchini et al., 2016; Bui et al., 2012; IGOS, 2004; Shahabi et al., 2014; Wang et al., 2016). While landslide occurrences include 9% of all natural disasters in the past decade, it is expected that this trend will increase in the coming years, due to the development of urbanization, deforestation and climate change (Yilmaz, 2009; Zare et al., 2013). The damaging effects of landslides include loss of life, rapid soil loss, and degradation of agricultural lands, gardens, roads, and engineering structures (Hassanzadeh Nafuti et al., 2012). Given the extent of the damages mentioned, it was explicitly stated that the cost of studying this phenomenon is much less than the damage. Therefore, to understand the susceptibility of hill slopes, landslide risk zones in different regions are addressed (Shadfar et al., 2007). Landslide susceptibility has been defined as the probability of a landslide occurring in a region based on local terrain conditions (Brabb, 1984; Ciampalini et al., 2016). Zoning and preparation of a landslide susceptibility map is a complex process (Brabb, 1991; Chen et al., 2016) that shows possible and sensitive areas to landslides through some effective factors by generalizing the occurrence of slope failures (Akgun, 2012; Van Westen, 2000). Landslide susceptibility maps provide important and valuable information for predicting landslides hazards which include an indication of the time scale within which particular landslides are likely to occur in the future (Atkinson and Massari, 2011). In light of GIS based landslide risk mapping, the multicriteria decision analysis (MCDA) methods provides a rich collection of procedures and techniques for structuring decision problems and designing, evaluating and prioritizing alternative decisions (Feizizadeh and Blaschke, 2014; Feizizadeh and Kienberger, 2017). MCDA has been widely applied to support environmental planning processes, where MCDA can provide a transparent combination of a problem from different perspectives and a systematic assessment of the alternatives (Huang et al., 2011; Keisler and Linkov, 2014; Kiker et al., 2005; Mustajoki and Marttunen, 2017; Voinov et al., 2016). There has been a vast body of research around the world on evaluation of landslide mapping based on GIS-MCDA methods (Feizizadeh and Blaschke, 2013; Feizizadeh et al., 2014a; Feizizadeh et al., 2014b). The GIS-MCDA methods and models used by researchers to prepare a landslide risk map are the analytical network process (ANP) (Abedi Gheshlaghi and Valizadeh Kamran, 2016; Neaupane et al., 2008; Neaupane and Piantanakulchai, 2006; Roostaei et al., 2015), fuzzy methods (Anbalagan et al., 2015; Bibi et al., 2016; Bui et al., 2015; Pourghasemi et al., 2012; Tangestani, 2009; Vakhshoori and Zare, 2016), neuro-fuzzy hybrid methods (Aghdam et al., 2016; Dehnavi et al., 2015; Pradhan, 2013; Vahidnia et al., 2010), and logistic regression (Ayalew and Yamagishi, 2005; Bui et al., 2016; Demir et al., 2015; Devkota et al., 2013; Sangchini et al., 2016; Umar et al., 2014).

Industrial companies are continually looking to improve their business turnover and improve practices to achieve this objective. Vincent et al. (2013) analyzed overall performance models to allow the decision-makers to select potential actions to launch, regarding given decision criteria. The Multi Criteria Decision Making (MCDM) approaches have been developed to deal with problems such as ranking, sorting and quantification. Gurumurthy and Kodali (2012) identified the different attributes that are to be considered when making the decision to select a suitable product development methodology. For the selection of such a methodology, especially for making a strategic decision within the product design and development department of an organization, the most commonly used MCDM model is the Analytical Hierarchy Process (AHP). Goyal et al. (2015) identified and prioritized corporate sustainability practices to improve the corporate sustainability performance in the manufacturing sector. Further, these practices were being prioritized to discover the essential practices to ensure logical allocation of limited resources. Then, AHP was used to assess the identified practices of corporate sustainability and to find their priorities for improvement of corporate sustainability performance. Globalization has opened up new opportunities and challenges for companies and brands (Alden et al., 1999). In the past, globalization of a product was a sign of its quality and the success of the company that produced it. Nowadays, products are often globally available irrespective of their reception and quality, and customers have a chance to assess and compare the quality of products from local and international producers. In this process, a local brand may have a chance to fulfill the requirements of the customers and win their satisfaction, while a well-established global brand may fail to do so (Milberg and Sinn, 2008). Globalization exposes customers in most countries to many foreign and local brands. Increasing competition has occurred between local and foreign brands not only in developed countries (e.g. Japan) but also in countries with newly industrialized economies (e.g. Korea and Mexico), as well as developing countries such as Iran. In this mass market, retail talent is considered a valuable resource in branding a product, and the service quality presented by retailers will add to a brand’s reputation. It is possible for companies to place themselves in the mass market by providing interactions with important customers who develop a brand’s values and advantages. A new trend in marketing science focuses on marketing activities in order to attract the attention and reception of local customers, and to be remembered as a quality product by them. Different studies show that the perception of potential customers of the value of a product is largely independent of its real value, i.e., the quality of the product or the services provided to the customers, rather, it relies primarily on success in branding and marketing activities (Kapferer, 2008; Keller, 2008; Kim et al., 2015).

Energy is an indispensable resource for all human activities—used in different facets of life such as cooking, heating, cooling, lighting, transportation, and production/manufacturing. Energy projections have shown that the need for energy has grown significantly for the last couple of decades, and is expected to maintain such an increasing trend. According to Energy Industry Agency, world energy consumption will increase 56% by 2040 [13]. Increasing demand and the ever so expanding gap between energy production and consumption could lead to global energy crisis, if not dealt properly in a timely manner. Increasing population, industrialization and urbanization force governments to re-make strategic decisions to meet the increasing energy demands. Many countries, including Turkey, have developed energy action plans to address the issue and to mitigate the potential impact of energy shortages. Energy consumption of most every country on the face of the earth has been dependent on the fossil fuels that include petroleum, coal and natural gas, but the reserves of fossil fuels are limited and nonrenewable (this is true even if we consider the latest developments in the oil and gas production industry—i.e., fracking). It is a common belief that in order to sustain daily lives in a reliable way, we need to minimize the usage rate of fossil fuels, since fossil fuels damage the ecosystem and can cause and accelerate the global warming phenomenon. Because of this, renewable energy and related sources/procedures have been gaining increasingly more attention. Awareness of environmental issues, depleting fossil fuels resources, the precarious nature of dependency on fossil fuel/oil imports laid the foundation for the recent interest in the exploitation of renewable energy sources [19]. Renewable energy is deemed to be clean, sustainable, cost-effective, reliable and environmental friendly, and hence can be relied on for the long-term.

مقاله سال 2017

در مدیریت زنجیره تأمین، انتخاب تأمین‌کننده می تواند به عنوان نوعی از مسأله های تصمیم گیری چند معیاری در نظر گرفته شود زیرا با معیار های مختلف درگیر است و ساختار سلسله‌مراتبی اغلب اوقات وجود دارد. این مقاله یک نوع از مسأله های تصمیم گیری چند معیاری را با معیار های دوترازی بررسی می کند که یک روش تجمعی ترکیبی جدید از TL-ANP (فرایندهای شبکه ای تحلیلی زبانی دوتایی) و IT-ELECTRE II (حذف دوتایی بازه و واقعیت انتقال انتخاب II) است. با در نظر گرفتن برهم کنش بین معیار ها، یک رهیافت TL-ANP که در آن ماتریس های مقایسه، روابط اولویت زبانی دوتایی سازگار هستند برای تعیین وزن معیار ها و زیر معیار ها به کار برده شده است. برای تعامل با موردی که معیارها همدیگر را جبران نمی کنند، یک رهیافت IT-ELECTRE I پیشنهاد شده است. در این رهیافت، امتیازدهی های انتخاب ها در مورد زیر معیار به صورت متغیرهای زبانی دوتایی بازه نمایش داده می شوند. یک درجه امکان و اولیت مبتنی بر احتمال  برای مجموعه های تطابق، عدم تطابق و بیتفاوتی تعریف شده است. سپس، شاخص های تطابق و عدم تطابق شناسایی شده و به کار برده می شوند تا شاخص های تطابق خالص و عدم تطابق خالص تشکیل شوند. در ادامه، مقادیر غلبه جامع از انتخاب ها برای رتبه‌بندی انتخاب ها حاصل می شود. در نتیجه، یک روش هیبریدی جدید برای تصمیم گیری چندمعیاری با معیار های دوترازی در محیط زبانی دوتایی بازه به دست می آید. در انتها، یک مورد واقعی انتخاب تأمین‌کننده بررسی می شود و تحلیل های مقایسه ای برای نمایش کاربرد و برتری روش پیشنهادی در دستور کار قرار می گیرد.