Research On Key Problems Of Reconfigurable Manufacturing System Control Framework Based On Entity Physical Behavior Planning

Today,returning to manufacturing has become one of the important strategies for the world's major countries,such as Germany's "Industry 4.0" program and the US "Industrial Internet" program.In China,the government has proposed the strategic outline of “Made in China 2025”,which specifies a manufacturing evaluation system consisting of 22 indicators,where the three indicators,i.e.,per capita manufacturing added value,full labor productivity and sales profit rate,take up a higher weight and reflect the importance of improving manufacturing efficiency in the outline.In order to improve manufacturing efficiency and reduce production downtime,researchers have proposed a device paradigm called Reconfigurable Manufacturing System(RMS).The smart card personalization system is an automated system that satisfies the RMS paradigm.Due to the reasons of the customer needs and machine characteristics,smart card personalization manufacturing industry puts forward some new requirements for the control system and its development process.Although there have been many achievements in the field of automated system control,especially in the literature on RMS control,they do not fully meet the needs of smart card personalization manufacturing yet.The existing results have the following problems:1.Unsupportive for mass individualization: The smart cards to be produced in the machine may undergo different production processes and transmission paths separately.The control system must be adapted to this mass individualization.However,the existing RMS controllers do not support production control scenarios for multi-material and multi-processes situation.2.Only supportive for data-flow-level integration: the development and maintenance of mechatronic systems are based on the integration of the module physical behavior which the control system must be adapted to.However,the existing machine description method describes the behaviors of the modules only through the interaction of the data flow among the module controllers,rather than the entity physical activities in the modules,and the availability of the behavior description is insufficient.3.Little support for legacy devices: The control system must be able to implement control integration for modules with restrained and limited access modes.However,in the existing research achievements,the field-level control implementation at the bottom of the framework is mostly bound to a single technology platform and lacks sufficient compatibility for integrating legacy devices.4.Low coverability on the scope of reliability: Since the behavior of the device is automatically generated by the control system,the behavior must be proven to be reliable and given the scope of reliability.Although there are many proven methods for generating reliability,due to the domain specificity,the scope of reliability cannot cover the scope of discrete manufacturing systems.5.Insufficient usability: In the face of complex control,the system is required to automatically infer the control strategy according to the requirements submitted by the mechatronic engineers,so as to avoid error-prone and time-consuming manual work such as designing strategies and writing software,and to improve R&D efficiency.However,most of the requirements description and reasoning methods belong to the field of computer science,which is not conducive for the mechatronic engineers to adopt.Taking all the aforementioned problems into consideration,this dissertation proposes a reconfigurable manufacturing system control framework based on entity physical behavior planning,which consists of three parts: control architecture,module behavior description and automatic reasoning.First,this dissertation proposes an RMS controller architecture that supports mass individualization production and is compatible with legacy devices.Inspired by the software design pattern "request-response",each material to be processed sends requests to the controller,and waits for a response via the device execution.This design pattern is suitable for scenarios of multi-material and multi-processes situation.In terms of control implementation,thanks to the mapping mechanism between the module's executable instructions and physical behavior of mechatronic module entity,the portability of the architecture among different field-level control implementation platforms is enhanced.Secondly,this thesis proposes a device module description method based on entity physical behavior.Unlike simulation systems,this approach is a simplified description of the physical world.Based on this description method,a definition on machine behavior safety constraints(material processing safety conditions)and material processing requests is given.This method solves the problem that the mechatronic engineers and software engineers cannot use the common language to accurately describe the module behavior specification and to coordinate the system development due to the lack of a suitable description language.Based on the description of physical behavior of module entities,the essence of system behavior is effectively grasped,and the unification of description on related domain knowledge is realized,which provides a basis for control reasoning automation.Thirdly,this dissertation presents a control strategy reasoning method based on automated planning and instruction mapping.This method transforms the strategic reasoning problem into an automated planning problem,and gives the corresponding automated planning problem definition.Based on the definition and the off-the-shelf planner,a serial solution planning method that supports the expression of the concurrent activities of the entities is implemented.At the same time,the proof of satisfiability is given to prove that any solution to the above automated planning problem satisfies the material processing safety conditions.In the aspect of instruction mapping,a semantic description of module executable instructions and an algorithm for mapping entity behavior to instruction sequences are proposed.Combined with the automated planning and mapping algorithm,the reasoning method can reduce the manual operation and ensure the correctness and security of the system behavior.Lastly,this thesis presents an example of a control application for a large smart card personalization manufacturing machine.According to the material spatio-temporal relationship in the pipeline,the request input set of the example is given,and the experiment of the control strategy reasoning is carried out for all the inputs.The relevant experimental results are obtained,and the practicality of the method in the actual industrial system control is demonstrated.In summary,the RMS control framework proposed in this thesis can effectively meet the control requirements of the smart card personalization manufacturing system.The main contributions of this paper are to solve the key problems that have long plagued the collaborative development of mechatronic engineers and software engineers by breaking through the boundaries of their respective professions,and to realize the formalization of requirement description.On this basis,by introducing control reasoning automation,it improves the quality of system research and development,reduces the uncertainty of the research and development progress caused by manual work,and ultimately achieves the purpose of improving the production efficiency of both the manufacturing equipment industry and the manufacturing industry.