网络物理自动化系统的虚拟工程外文文献

Contents lists available at ScienceDirect Advanced Engineering Informatics journal homepage Full length article Virtual engineering of cyber physical automation systems The case of control logic Georg Ferdinand Schneidera c Hendro Wicaksonob c Jivka Ovtcharovac aFraunhofer Institute for Building Physics IBP and Technische Hochschule N rnberg F rther Stra e 250 90429 N rnberg Germany bJacobs University Bremen Campus Ring 1 28759 Bremen Germany cKarlsruhe Institute of Technology KIT Institute for Information Management in Engineering Kriegsstra e 77 76133 Karlsruhe Germany A R T I C L E I N F O Keywords Control logic Cyber physical systems Ontology Industrial automation Virtual engineering A B S T R A C T Mastering the fusion of information and communication technologies with physical systems to cyber physical automation systems is of main concern to engineers in the industrial automation domain The engineering of these systems is challenging as their distributed nature and the heterogeneity of stakeholders and tools involved in their engineering contradict the need for the simultaneous engineering of their cyber and physical parts over their life cycle This paper presents a novel approach based on the virtual engineering method which provides support for the simultaneous engineering of the cyber and physical parts of automation systems The approach extends and integrates the life cycle centered view mandated by current conceptual architectures and the digital twin paradigm with an integrated iterative engineering method The benefits of the approach are highlighted in a case study related to the engineering of the control logic of a cyber physical automation system originating from the process engineering domain We describe for the first time a modular domain ontology which formally describes the cyber and physical part of the system We present cyber services built on top of the ontology layer which allow to automatically verify different control logic types and simultaneously verify cyber and physical parts of the system in an incremental manner 1 Introduction The coalescence of computation and physical processes through networked embedded devices is recognized as Cyber Physical Systems CPS 1 These systems are identified across various domains and their successful deployment is considered as a remedy to global chal lenges for engineers in different industries such as automotive health care energy and industrial automation 2 In particular systems from the industrial automation domain qualify as CPS 3 4 as they consist of distributed and networked embedded devices where a cyber part e g the control of a robot or process is combined with physical parts such as robots or conveyors transporting goods As a long history of research exists specifically addressing the needs of industrial automa tion systems before considering them as CPS we refer to this speciali zation of CPS as Cyber Physical Automation Systems CPAS The engineering of CPAS poses many challenges towards the in volved design and engineering teams 4 To successfully design CPAS interdisciplinary and often cross enterprise teams are needed which impinge with the heterogeneous and distributed nature of these systems 4 In addition to their inherent heterogeneity the existence of a vast number of engineering tools and their associated data formats for ex change of information among these tools challenge the seamless col laboration of the involved stakeholders over the life cycle of these systems 5 A number of problems and challenges are recognized when dealing with the engineering of CPS where different design techniques are investigated in this regard 6 A main demand is to find unified ways 1 to perform cross domain i e simultaneous design analysis and verification of both the cyber and physical part of CPS 2 6 i e Ty pically a particular formalism represents either the cyber or the phy sical process well but not both 7 CPS can be structured into their physical platform and cyber part 8 and Feedback loops 1 comprise a significant share of the cyber part of CPS This is a fundamental capability and concern in the en gineering of CPAS where some behaviour processes information ob tained from inputs to determine respective outputs for effectuating on a physical system under control 9 In literature different terms are used synonymically for this behaviour e g among others application logic 10 control logic 11 and software 12 Throughout this work we use the term control logic https doi org 10 1016 j aei 2018 11 009 Received 19 February 2018 Received in revised form 13 November 2018 Accepted 30 November 2018 Corresponding author at Fraunhofer Institute for Building Physics IBP and Technische Hochschule N rnberg F rther Stra e 250 90429 N rnberg Germany E mail address Georg Schneider ibp fraunhofer de G F Schneider Advanced Engineering Informatics 39 2019 127 143 Available online 27 December 2018 1474 0346 2018 Elsevier Ltd All rights reserved T Specific contributions analyzing the problems and challenges re lated to the engineering of control logic in CPAS exist and the current state of the art is summarized in Vyatkin 13 as well as current chal lenges and future research directions are identified by Vogel Heuser et al 12 14 Despite control logic being essentially a significant part of the computation component of CPS 8 the cited reviews 12 14 discuss the topic without relating to CPS in general However they conclude in agreement with the contributions related to general CPS 1 6 7 that the mechanical electrical and control logic domains need to be considered simultaneously In addition automated formal verification methods are demanded 12 14 to orchestrate the integration of separately developed software modules as well as to ensure their safety and reliability It is criticized that verification techniques often are only applied during early design stages and are restricted to one specific type of control logic 12 The proposed methods and tools for formal verification 12 14 constitute isolated solutions working only for some type of control logic and without simultaneously considering the cyber and the physical parts Typically multiple stakeholders are involved in the engineering process and use tools with associated disparate data formats impeding in formation exchange 15 The use of model based engineering methods is widely adopted during the design phase 12 However most model based engineering methods implement a forward oriented information flow thus the manual unavoidable changes on code level on the target Programmable Logic Controller PLC are not fed back to the model 12 A number of integrated engineering methods are proposed by scholars see Section 2 which address some of the mentioned pro blems However a missing ability is to enable an iterative holistic life cycle centered view on all knowledge associated to a respective CPAS generated during its requirement definition design commissioning and operation This view is intensively discussed in the industrial automa tion domain Here the availability of a structured semantically well defined collection of all digital artifacts related to a CPAS including among others engineering knowledge simulation models and opera tional data is envisioned to be collected following the digital twin paradigm 16 17 This paradigm is reflected by the envisioned con ceptual architectures of CPAS 18 where the interplay of the real CPAS and its virtual counterpart enables a whole new range of appli cations in the life cycle of a system 18 Moreover the digital twin is iteratively updated and verified along the life cycle Hence engineering methods need to both address the demands set by scholars to allow the simultaneous engineering of cyber and physical parts of CPS as well as support a life cycle centered view in the context of the digital twin paradigm and conceptual architectures of CPAS In essence the following problems can be identified Despite a disconnection between the research areas related to CPS engineering and control logic in industrial automation engineering a consensus exists in both areas that the cyber and the physical part need to be considered simultaneously e g in automated verifica tion The missing support of integrated engineering methods for an iterative holistic life cycle centered view on all knowledge across domains including potential reiterations particularly relating to the digital twin paradigm and corresponding conceptual architectures The analysis of existing contributions related to integrated en gineering methods automated verification methods existing data for mats and ontologies as presented in Section 2 shows that none of the existing works fulfill the following requirements to overcome the stated problems R 1 Need for an iterative holistic life cycle centered engineering method which particularly integrates well with the contemporary digital twin paradigm 16 and conceptual architectures of future CPAS 18 R 2 Definition of a formal model which integrates knowledge on the cyber and physical domains using a machine interpretable lan guage R 3 Integrate prevalent heterogeneous engineering tools and data formats R 4 Support of simultaneous engineering of CPS such as automated verification of both the cyber and the physical part R 5 Automated formal verification of control logic designs including different types of control logic 12 R 6 Incremental verification at changes and updates 12 R 7 Bidirectional flow of information from model to target format and vice versa 12 To overcome the shortcomings of existing work reviewed in Section 2 we propose instead a novel approach based on the Virtual En gineering VE method 19 20 We do so by integrating the iterative holistic life cycle centered approach of the VE method with a con ceptual architecture of CPAS in the industrial automation domain 18 and consider its relationship to the digital twin paradigm 17 We study the implications of our approach in a case study related to the engineering of a real world laboratory scale CPAS originating from process engineering domain 21 Core foundation of this approach is the definition of a common semantic integration layer We present a novel domain model which integrates both the control logic cyber part and the plant physical part of a CPAS by means of ontology This work builds on and extends a previous contribution presenting models for the explicit formal modeling of control logic in automation systems 11 and extends it into the area of CPAS To address the prevalent heterogeneity in engineering tool chains our solution is based on Se mantic Web Technologies SWT as these provide means to integrate heterogeneous formats into one semantic integration layer 5 22 The domain model is implemented using the Web Ontology Language OWL 23 which constitutes an expressive and machine interpretable language due to its support for reasoning and semantic search 24 The novel domain model allows to perform simultaneous engineering tasks of CPAS e g within the case study we detail the simultaneous auto mated verification of control logic and plant information Moreover we highlight the capabilities of our approach in enabling automated formal verification across different types of control logic incremental ver ification of changes in the design and the support of a bidirectional flow of information from and to the common semantic integration layer The remainder of this paper is structured as follows In Section 2 we analyze existing work on their ability to fulfill the above stated re quirements with a specific focus on control logic engineering in CPAS Section 3 then highlights how the iterative holistic life cycle centered approach of the VE method integrates particularly well with the con ceptual architectures for CPAS and the digital twin paradigm Finally we detail the usage of our approach in a case study related to the en gineering of a CPAS from process engineering in Section 4 The case study includes in Section 4 2 the definition of a formal domain model for both the cyber and physical part of the studied CPAS and a scenario based evaluation of formal automated verification tasks in Sections 4 3 4 5 2 Related work A comprehensive body of research related to the engineering of CPS and CPAS exists To organize related contributions we consider an il lustrative engineering work flow as established in most current model based engineering methodologies which is shown in Fig 1 We analyze contributions related to integrated engineering methods for general CPS and CPAS in their ability to fulfill the respective needs arising from the problems as specified above Second as formal verification is con sidered as one of the most important tasks related to CPS engineering we review formal verification methods with respect to the formulated G F Schneider et al Advanced Engineering Informatics 39 2019 127 143 128 requirements Finally in the need of formal models for the respective domains we analyze existing data formats and ontologies in their ability to serve as a domain model to enable automated engineering methods and support We do not review specific tools for specification design and execution as these are use case and domain specific Some of the tools used in the case study presented in Section 4 are referenced in Fig 1 2 1 Integrated engineering methods The engineering of CPS is a complex task due to their inherent heterogeneity of cyber and physical components For CPS in general Khaitan VE2 Creation For physical platform and computation components of a CPAS abstractions exist 8 Based on these abstractions different software tools export this knowledge to various formats Adapters extract transform and load this input to the information integra tion and storage layer thus populating the knowledge base with case specific knowledge VE3 Virtualization In the cyber service layer simulation models and virtual builds for e g virtual commissioning are generated and functionality can be tested and verified against requirements Model software and hardware in the loop testing methods are deployed for testing and validation VE4 Validation Methods for formal verification are available as a ser vice to test and prove safety and reliability Code quality is en sured by continuous integration and automated testing VE5 Analysis The design decisions and choices need to be analyzed and thoroughly tested through virtual mock up units simulation based testing and virtual commissioning VE6 Review and continuous improvement The VE method is an iterative process and requires continuous improvement in multiple iteration cycles through reviews and conferencing In the next section we further detail and present the successful de ployment of the method to a case study related to a use case from process engineering 4 Case study virtual engineering of the experimental batch plant Within this section we study the implications of our approach in a case study related to the engineering of a real world laboratory scale CPAS originating from process engineering domain The specific per spective of the discussion is related to the engineering of control logic in CPAS however the physical parts are considered as well In Section 4 1 we provide background information on the chosen CPAS and show how we reuse the existing engineering documentation tables drawings text modelica source code 21 61 as a starting point for our study Next we present the results from subsequently executing the steps of the VE method We present a novel ontology based modular domain model in Section 4 2 which explicitly and formally describes both cyber and physical plant domains It extends the work initiated in Schneider et al 11 Then in Section 4 3 we highlight the capabilities of our approach in integrating heterogeneous tools and associated data formats of a control logic engineering tool chain to the novel modular domain model In Section 4 4 we describe how our approach enables automated verification tasks to support the simultaneous engineering of cyber and physical components of a CPAS Finally we highlight the support of the method for bidirectional flow of information and incre mental verification in Section 4 5 4 1 Description For our use case we study a real world laboratory scale CPAS The Experimental Batch Plant 21 The plant realizes a batch process for the desalination of water and originates in the process engineering domain It qualifies as a CPS as it involves a physical part e g tanks pipes heaters fluids etc and a cyber part e g the control logic run ning on a PLC Both parts are connected via sensors and actuators of a computer network to realize the automatic control of the process We choose the plant as it is designed to serve as an open use case for re searchers with a reasonable amount of complexity with regard to the plant physics and cyber part involved as well as simplicity of the process which helps the readers to understand rapidly The engineering of the system is documented in tables textual de scriptions and drawings by Kowaleski et al 21 Moreover from an additional effort 61 an open source simulation model implemented in Modelica 62 is available As illustrated in Fig 3 we use this doc umentation as an input to our case study and design and implement an engineering tool chain which we integrate with our proposed approach for the virtual engineering of CPAS The initial documentation of the plant 21 includes a thorough textual description of the batch process including tables with e g the sizing of the physical components a Piping and Instrumentation Dia gram PI next the semantics of sensors ac tuators and control actors and finally the domain of the plant under control The nexus in the modular ontology structure see Fig 6 is the CTRLont ontology 11 which formally describes the relationships betweens explicit models of control logic and adjacent domains such as sensors actuators and plants To specify units and quantities withi