As a promising service-oriented manufacturing paradigm, CMfg has attracted attention from many scholars and extensive investigations have been done. Full-scale sharing, free circulation, and optimal collaboration of manufacturing resources and capabilities are key issues in CMfg. Tao et al. investigated the applications of IoT technologies in CMfg to achieve intelligent perception and access of manufacturing resources [10]. Additionally, combing the CC technology, they proposed a CC-based and IoT-based CMfg service system and its architecture [11]. Resource virtualization and encapsulation are critical for CMfg. Morariu et al. introduced the virtualized MES and shop floor architecture in a private cloud to reduce operational costs and improve the flexibility, agility, and maintainability of the manufacturing system [12]. Liu et al. analyzed the features of resources and constructed a multilevel resource virtualization framework [13]. To achieve the formal description of the manufacturing capabilities in CMfg, Luo et al. established a multidimensional information model of manufacturing capabilities [14]. Xu et al. studied the dynamic modeling of manufacturing equipment capability by establishing the mapping relationship between real-time condition data and the ontology model of equipment capability [15]. Based on the IoT and CC technologies, Zhang et al. presented a service encapsulation and virtualization access model for manufacturing machines [16]. After cloud services published into the CMfg platform, how to realize the efficient discovery of services is another hotspot. Guo et al. proposed an agent-based service discovery framework, within which the task agent and service agent are included. And a structural matching method based on both static and dynamic parameters was studied [17]. Li et al. adopted the Ontology Web Language (OWL) to describe the manufacturing services and based on that a similarity algorithm is proposed. Then, a five-step service matching process has shown its advantages [18]. In the CMfg environment, SMEs are typical participants. Huang et al. discussed the manufacturing resource and capability sharing for SMEs and proposed an SME-oriented CMfg service platform, of which the architecture and key technologies are studied in details [19]. Song et al. designed a CMfg service platform for SMEs and researched some common engines like the intelligent matching engine to support a series of manufacturing business processes in CMfg [20]. Wu et al. applied the concept of CMfg in semiconductor manufacturing operations and proposed a semiconductor industry-oriented architecture for CMfg. Finally, a case study showed the created values for the customers and suppliers [21]. Yang et al. studied how federal resources collaboratively complete large complex projects in CMfg mode. And a large equipment complete service (LECS) collaborative logical framework based on CMfg platform and generalized partial global planning (GPGP) is constructed [22]. With the generation of massive, complex data from the RFID-enabled shop floor in CMfg environment, it is more difficult to extract and process data to make them meaningful. Zhong et al. used a RFID-Cuboid model to reconstruct the raw data according to production logic and time series [23].

The IoT is considered as a part of the Internet of the future and will consist of billions of intelligent communicating “things” [24]. In recent years, IoT technologies such as RFID, sensors and so on have been widely applied in the industrial field and significant contributions have been made. In the shop floor environment, physical manufacturing objects are equipped with RFID devices to become “smart,” then the real-time production data can be collected [25]. This promotes the real-time traceability, visibility, and interoperability in shop floor planning, execution, and control [26], enhance the implementation of advanced manufacturing strategies and technologies as well [27]. Yang et al. developed an online sequential extreme learning machine (OS-ELM)-based RFID positioning method in the manufacturing execution system, using RFID to efficiently acquire the real-time data of manufacturing objects (MOs) and the OS-ELM to support real-time data processing [28]. With the support of IoT, Zhang et al. presented a real-time information capture and integration architecture of the Internet of manufacturing things (IoMT) [29]. They designed a ubiquitous shop-floor environment powered by wireless devices like RFID. Then, a framework of multiagent-based real-time production scheduling is proposed to achieve the close loop of production planning and control [30]. To realize the collection and synchronization of the real-time data, Luo et al. created a ubiquitous manufacturing (UM) environment in a hybrid shop floor [31]. Zhong et al. developed an advanced production planning and scheduling (APPS) model for the RFID-enabled real-time ubiquitous environment. The model was tested through four dimensions. They found that the release strategy based on real-time information can reduce the total tardiness effectively and the model was immune to disturbances like defects [32]. Guo et al. proposed an RFID-based intelligent decision support system for real-time production monitoring and scheduling in distributed manufacturing environment. This system was verified to enhance the production efficiency, reduce the production waste and the labor cost as well [33]. Arkan et al. presented a RFID-based real-time location system (RTLS) to obtain work-in-process visibility in manufacturing based on the real-time shop-floor data [34]. Zhang et al. designed an optimization method for shop-floor material handling. And through a case study, the effectiveness of the method was analyzed considering the empty-loading ratio and total distance [35]. Zhong et al. proposed a big data approach to mine trajectory knowledge from the RFID-enabled shop-floor logistics data to support decision makings like logistics planning and scheduling [36]. Qu et al. presented an IoT-enabled real-time production logistics (PL) synchronization system to deal with dynamics occurring in the PL processes [37]. Wang described the significance to implement RFID technology into Norwegian manufacturing industry and developed an intelligent and integrated RFID system to improve traceability and visibility in manufacturing process [38].

In the CMfg environment, how to optimally select services from massive candidate services; then how to orchestrate the service compositions, are two critical problems. In fact, the selection of service compositions is far more complex than that of single services. Service composition and optimal selection (SCOS) is a typical multiobjective combinatorial optimization problem [39], Tao et al. investigated the problem with multiple objectives and constraints in CMfg and developed a parallel intelligent algorithm to solve it [40]. Huang et al. established the categories of services and respective quality of service (QoS) indexes for SCOS in CMfg and designed a new chaos control optimal algorithm (CCOA) [41]. Liu et al. proposed a “Multicomposition for each task” (MCET) pattern that integrates the incompetent composite services to execute multifunctionality manufacturing tasks. And a hybrid operator–based matrix-coded genetic algorithm (HO-MCGA) is designed [42]. Liu et al. proposed a synergistic elementary service group–based service composition (SESG-SC) to relax the one-to-one mapping between services and subtasks to enhance the overall QoS level and success rate of service composition [43]. Xiang et al. developed a multiobjective optimization algorithm combining the group leader algorithm (GLA) and the idea of Pareto solution for the SCOS problem based on QoS and energy consumption [44]. Xue et al. proposed a QoS model of the service composition considering the horizontal collaboration between services in cluster supply chains (CSC) and designed a genetic-artificial bee colony (G-ABC) algorithm to efficiently find the optimal solution [45]. Zheng et al. presented a design preference-based QoS evaluation method for resource service selection and adopted the particle swarm optimization (PSO) algorithm to select the optimal service composition [46]. The correlations among services may affect the aggregation QoS level of service compositions. Tao et al. took service correlation into consideration for the multiobjective MGrid resource service composition and optimal-selection (MO-MRSCOS) and proposed a PSO-based method for solving this problem [47]. Jin et al. built a correlation-aware service description model to characterize the QoS dependence among services. Based on this, they presented a service correlation model to automatically obtain QoS values of services. Finally, the correlation-aware service optimal selection is solved by the genetic algorithm (GA) [48]. Transportation is another factor influences the QoS property of service composition, which primarily impacts the total service time and cost [49]. Lartigau et al. analyzed the geoperspective correlation among services and developed an adapted Artificial Bee Colony (ABC) algorithm with an initialization enhancement to optimize the computational time [50]. Xiang et al. introduced the “Big manufacturing data” in CMfg, which have brought difficulties and challenges to the SCOS problem. For solving this problem, they presented a case library-based initialization method for the optimization algorithm [51].

In this section, we studied the information model of cloud machine with both its static description information and dynamic manufacturing information considered. As shown in Fig. 5.2, the cloud machine model comprises four kinds of information, including the basic information, the function information, the real-time status information, and the evaluation information. It is defined as:

The efficiency and QoS discovery and service matching greatly depend on how to describe the manufacturing services. In this work, ontology and semantic web technologies are employed for achieving aforementioned goals. The semantic web provides a common framework that enables the sharing and reuse of data across the applications, enterprises, and communities [52]. Ontologies can not only explicitly represent the domain knowledge and clarify their relationships but also have strong reasoning ability. To effectively express the manufacturing information as well as the connotative meanings, the ontology description language OWL-S is utilized to describe the ontology model. As OWL-S can semantically describe the web services according to the capabilities offered and perform logic inference to service match between the offered capabilities and the required capabilities [53].

Universal description, discovery and integration (UDDI) is a specification for web-based information registries of web service. It is also a publicly accessible set of implementations of the specification that define a way for businesses to register and publish the web services they provide, so that other businesses can discover them [54].

Based on the UDDI technology, we built a MS-UDDI to realize efficient and integrated management of manufacturing services; to allow service providers to register and publish their manufacturing services into the CMfg platform. Therefore, these services can be easily found and invoked by demanders. As shown in Fig. 5.5, the framework of MS-UDDI consists of three submodules, including the registration module, the publishing module, and the search module.

In today’s manufacturing industry, distributed SMEs are insufficient to find tasks proactively, which results in the low utilization of manufacturing resources. To improve this, we presented a task-driven service proactive discovery mechanism that enables CMSs to actively make rapid responses to tasks and then apply to perform corresponding ones. Once applications are made, a semantic-based intelligent match from the functional perspective is performed between CMSs and tasks to select competent services. As shown in Fig. 5.7, the semantic matching is performed considering all the functional attributes of CMSs.

CMSs in the candidate set offer the same kind of functional services to a specific demander but still differ in many other service characteristics like cost, time, and so on. How to choose a CMS to best satisfy the demander’s customized requirements is studied in this section. A service optimal configuration method is proposed aiming to find an optimal solution for the task. The specific procedure is shown in Fig. 5.8, in which an evaluation system and a GRA-based evaluation approach are included.