Abstract
The reconfiguration manufacturing system (RMS) allows mass manufacturing to be changed into customized mass production and short-series production. However, the existing RMS models are not applicable to the printing production process, which is coexistence of the hybrid nature of being continuous and discrete. With the printing manufacturing system as the object of the research, this paper proposes the Agent-Resource-Workstation (ARW) model based on agent by analyzing the characteristics of “Production and Consumption of Resources” in the printing process. And then, each component of the ARW model is analyzed and defined. Finally, after systematically abstracting the reconfiguration of workstation models (RWMs), the reconfigurable mechanism of RWMs is established based on the algebra of communicating process (ACP), and formal verification was conducted by applying axioms of process algebra. The verification results show that RWM can show its external behavior well. Meanwhile, our research provides a theoretical approach for the reconfiguration of the printing manufacturing system.
1. Instruction
At present, the direction of international advanced printing manufacturing system is going gradually to digital network and intelligence. Reconfigurable printing manufacturing system (RPMS) provides the best way to do this. In the late 1990s, Y. Koren, the professor of University of Michigan, proposed the concept of reconfigurable manufacturing system [1]. The model and design theory of reconfigurable systems mostly focused on the discrete manufacturing system, which is typical of machining. Because function block model is regarded as the key to improve the efficiency of reconstruction, we abstract model from the function block of system before reconstructing the printing manufacturing system, and then the model is formally verified. Finally, the reconfigurable principle of the manufacturing system is explored.
Recently, agent-based function block and IEC 61499-based function block are proposed for the RMS. Mulibika and Basson [2], as well as Kruger and Basson [3], have compared these two approaches. Although Valentea, Mazzolinib, and Carpanzanoa [4] have confirmed that IEC 61499 holds promise for RMS applications, there has been a severe lack of support for this standard by major automation controller vendors, and the development platforms to support this approach are not mature enough to be attractive to the industry. IEC 61499 further does not make provision for dynamic instantiation, which inhibits its ability to implement holonic control architectures [5]. However, most research in RMS has used the agent-based system, such as referred to in [6–13]. A. Maka et al. have used agent-based modeling in modeling decision-making systems, which allows more flexibility than standard central management systems [6]. The use of agent-based systems allows mass manufacturing to be changed into customized mass production and short-series production, as recently confirmed by R Cupek et al. [7]. The most important properties of an agent are the autonomy, intelligence, adaptation, and cooperation [11].
Nevertheless, most research assumes that the agent in the manufacturing execution system can understand the interactive message and the perceptual environment and can be reconstructed spontaneously by reasoning. This approach has a high demand for the agent reasoning ability, and it is difficult to realize this technology. Since there is no corresponding reconfigurable mode guidance, so that a large number of interactions between agents exist and the reconfigurable efficiency of system are low. An ARC-model-based (Agent model, Role model, and Characteristic model) method for realizing this kind of reconfiguration and a corresponding ARMES prototype were proposed by National CIMS Engineering Research Center, Tsinghua University [12]. The data structure based on the ARC model is not adapted to the data structure in the RPMS based on the “Production and Consumption of Resources” production process model. If we apply reconfigurable theory based on ARC model to guide the reconfiguration of printing manufacturing system, the coupling between the information flow and the control system of the system is enhanced, and the reconfiguration efficiency is reduced. Therefore, On the basis of studying the influence mechanism of structured control information data on RPMS, a new reconfigurable theory based on Agent-Resource-Workstation (ARW) model is proposed in this paper.
When manufacturing systems are reconfigurable, A. Youssef proposed that the maximum investment constraints of configurations and the space constraints should be satisfied [13, 14]. Lou Hongliang et al. used Graph Theory to research on the reconstruction strategy of RMS full-life cycle under the variety and batch changes of parts [15]. Under the deterministic market demand environment, Duan Jianguo et al. discussed that production drives the reconfiguration method of the RMS [16]. All of the research [13–16] is independent of the control quantity, and the discrete machining system is the object of study. Dai Yuru and Wang Jian have used the Component Hybrid Petri net method to reconfigure manufacturing modeling for continuous chlorine production system [17]. However, these models are not applicable to the printing production process, which is coexistence of the hybrid nature of continuous and discrete. Since the process algebra [18] has precise formal semantics and rigorous mathematical foundation, the subsystem can be combined into a large system by operator, and the system actions can be hidden as an internal activity by hiding the operator, and the model can be formal verification based on the algebraic axioms. Currently, the process algebra is more used in the field of software communications and less applied in the reconfigurable manufacturing system [19]. Based on the ARW component model, this paper utilizes process algebra to abstract the attributes and behavior characteristics of reconfigurable system. Furthermore, the algebraic model of RWMs is established, and then the formal verification of model is carried out.
2. Proposed ARW Model
Figure 1 shows us the printing process of a book. It is easy to see that the essence of the printing production manufacture is a constant process, in which old resources (raw materials) are consumed and then new resources are produced. When applying the “Production and Consumption of Resources” model to define the printing production manufacture process, we can regard each production process as a workstation of processing resource. Resources are consumed to manufacture new resources, which are used by the subsequent workstation. Thus, the workflow is constituted. The resources are defined as the various types of information resources (such as various agreements, rules, and control information) and as entity resources (such as paper, ink, and printing equipment) in the workflow. For example, with the control information, text and images are designed into original, which is consumed to manufacture the PDF file by the Made-up (such as Adobe InDesign, CorelDRAW, and FanTart), as shown in Figure 1. Therefore, the Agent-Resource-Workstation (ARW) model is proposed based on the functions, attributes, and behaviors of the manufacturing units. The ARW model is composed of Agent model, Resource model, and Workstation model, as shown in Figure 2.
3. Reconfiguration of the Workstation Model
In the ARW model, the Agent model is defined as a component kernel, which depicts the processing functions and the structure of the Agent type. The Agent model is expressed by internal processing functions. The Resource model is a set of various resources. The attributes of resource include the resource type, the interaction of the materials, and the behavior constraints of the manufacturing suitability. The Agent model and the Resource model are integrated in Workstation model, which is expressed by interface and behavior. The Resource model is an interface of the Workstation model. The interface is a poset (Partially Ordered Set) that depicts the coupling relationship between the resources. The behavior consists of an internal processing functions’ set which depicts the execution relationship between the Agent model and the Resource model, the interface element of the interactive protocol, and a set of messages received and sent. When the Agent model executes the Resource model, it means that the Agent model promises to implement the service functions and the production behaviors (Production and Consumption of Resources) of the Resource model defined. The Agent model complexes with the others to integrate into a new Workstation only through the interface (Resource model). The reconfigurable principle of Workstation models based on the ARW model is shown in Figure 3. Since the output resources of W1 are the same as the input resources of W2 (W3) in Figure 3, namely, interface matching, W1 and W2 (W3) can be reconfigured. This is a reconfigurable type called additional reconfiguration. Another type is replaceable reconfiguration. Due to the production process changing, W2 has to be replaced by W3 and then reconstructed with W1. For example, in the production of a book, saddle stitches are required when signatures are no more than six. The adhesive binding machine in Figure 1 just needs to be replaced into the saddle stitching machine. W1 and W4 cannot be reconfigured, because the output resources of W1 are different from the input resources of W4, namely, interface mismatching.
The Workstation model can be reconfigured with another Workstation model or more Workstation models by additional reconfiguration. Depending on the different dimensions of the integration, the new Workstation model will contain two or more component kernels, as shown in Figure 4.
4. Mathematical Model
To specify more complex systems, the algebra of communicating process (ACP) [20] specification framework offers constants, operators, recursion, etc. as follows:(1)A constant is called deadlock, which does not display any behavior, and the other constant is called the silent step, which represents a sequence of internal actions that can be eliminated from a process graph.(2)A unary operator is called encapsulation for sets of atomic actions, which rename all actions in into . And the other unary operator is called abstraction for sets of atomic actions, which rename all actions in into .(3)The binary operators (+), (·), (), (), and () are called the alternative, sequential, merge, left merge, and communication merge composition operators, respectively. The three parallel operators (), (), and () bind stronger than (+) and (·).
If is a guarded recursive specification and is called recursion variable in , then is called guarded linear recursion, which intuitively denotes the process that has to be substituted for in the solution for .
Resource elements from a finite set are initial input resources (R1). Resource elements from a finite set are the output resources (R2) of the Workstation one (W1), and they are also the input resources of the Workstation two (W2). Resource elements from a finite set are the output resources (R3) of A2.
First, as shown in Figure 5, W1 reads a resource from channel A, then this resource is consumed to produce resource by the agent of the W1. Second, the W1 sends the resource into the channel B. However, when the resource is lost or the channel B is corrupted, so that the resource that cannot be received by the W2 can be turned into an error message , which is sent into the channel B. Thirdly, the W2 receives the resource or the error message via channel B. If the W2 receives the error message or the resource that cannot be manufactured by the agent of the W2, then it sends the acknowledgement message or into the channel D. The W1 will send the (right) resource into the channel B again when it receives the acknowledgement message or . If the agent of the W2 can use the resource to produce the resource , the W2 sends the resource into channel C. Finally, the W2 sends the acknowledgement message to the W1 via the channel D.
Applying ACP, we can describe the states of the W1:
Applying ACP, we can describe the states of the W2:
A send and a read action of the message over the same channel (B or D) communicate with each other:for . All other communications between atomic actions result in .
5. Formal Verification
The desired concurrent system is obtained by putting S and R in parallel, encapsulating send and read actions over the internal channels B and D, and abstracting away from communication actions over these channels. That is, the RWMs are expressed by the process termwithThe process term can be expanded by using axioms of process algebra as follows:Applying unary encapsulation operator , (6) can be encapsulated and reduced. We can deriveSimilarly, these derivations are sketched as follows:
The recursive specification of the RWMs, consisting of the recursive equations for the recursion variables S, T, U, R, Q, and P, can be transformed into linear form by introducing extra recursion variables to represent for . For notational convenience, process terms are abbreviated to in this section. Owing to (I)-(VI), RSP yieldswhere denotes the linear recursive specificationAfter application of the abstraction operator to the process term , the loops of communication actions (between recursion variables X2 and X4, X2, and X5) become τ-loops, which can be eliminated by using CFAR.Applying axioms of process algebra and (11) we can deriveIn combination with (9) this yields
Equation (13) shows that the desired concurrent system can successfully read the resources and manufacture the resources to send it. It means that the RWMs exhibit the desired external behavior. The formal verification of the RWMs is finished.
The reconfiguration of the manufacturing system must meet the conditions of interface matching. It means that, in the reconfigurable printing manufacturing system, the output resources of each workstation must be produced by the subsequent workstation. Formal verification results show that the RWMs can realize the reconfiguration of the printing manufacturing system.
6. Conclusions
In this paper, reconfigurable mechanism of manufacturing system for printing process based on ACP is presented and studied. In the fast-varying production environment, the target of the printing manufacturing system can make the best printing process in time according to the technological requirements of product. The following work is mainly completed: Firstly, by analyzing the characteristics of “production and consumption of resources” in the printing process, this paper proposes the ARW model based on agent with the printing manufacturing system as the object of the research. Secondly, each component of the ARW model is analyzed and defined through the analysis of the printing equipment and materials. Finally, the reconfigurable mechanism of the RWMs is established based on ACP after systematically abstracting the RWMs, and formal verification was conducted by using axioms of process algebra. The verification results show that the RWMs can display its external behavior well. Meanwhile, the reconfigurable mechanism of the printing manufacturing system is sufficiently stable for the rapid change in the printing process.
Appendix
A. Axioms for BPA
A1 ; A2 ; A3 ; A4 ; A5 .
B. Axioms for PAP
M1 ; LM2 ; LM3 ; LM4 ; CM5 ; CM6 ; CM7 ; CM8 ; CM9 ; CM10 .
C. Axioms for ACP
A6 ; A7 ; D1 ; D2 ; D3 ; D4 ; D5 ; LM11 ; CM12 ; CM13 .
D. Recursive Definition and Specification Principles
RDP ; RSP , then .
E. Axioms for Projection Operators
PR1 ; PR2 ; PR3 ; PR4 ; PR5 .
F. Approximation Induction Principle
AIP .
G. Axioms for the Silent Step
B1 ; B2 .
H. Axioms for Abstraction Operators
TI1 ; TI2 ; TI3 ; TI4 ; TI5 .
I. Cluster Fair Abstraction Rule
CFAR If is in a cluster for with exits
Nomenclature
| : | Input resources of the W1 |
| : | Output resources of the W1 and input resources of the W2 |
| : | Output resources of the W2 |
| : | A set of resources |
| : | A set of resources |
| : | A set of resources |
| : | The acknowledgement message that the resource cannot be processed by the agent of the W2 |
| : | The error message |
| : | The acknowledgement message that can be processed by the agent of the W2 |
| : | Read the resources from the channel A |
| : | Read the resources from the channel B |
| : | Read the message from the channel D |
| : | Read the message ⊥ from the channel B |
| : | Read the message from the channel D |
| : | Send the resources into the channel B |
| : | Send the resources into the channel C |
| : | Send the message into the channel D |
| : | Send the message ⊥ into the channel B |
| : | Send the message into the channel D |
| : | The initial state of the W1 |
| : | The state of successfully manufacturing resources |
| : | The state of waiting the acknowledgement message |
| : | The initial state of the W2 |
| : | The state of receiving the error message |
| : | The state of successfully manufacturing resources . |
Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.
Acknowledgments
The authors would like to thank their colleagues from the Printing Technology Research Group for helpful discussions and comments on this paper. The project is supported by Natural Science Foundation of the Science and Technology Department of Shaanxi Province (Grant no. 2016JM5068), Natural Science Foundation of the Key Laboratory of Shaanxi Provincial Department of Education (Grant no. 15JS075), and Natural Science Basic Research Plan in Shaanxi Province of China (Grant no. 2017JM1028). The project is also supported by Shaanxi Collaborative Innovation Center of Green Intelligent Printing and Packaging.