Post on 18-Mar-2020
transcript
Entwurf, Simulation und Implementierung ereignisdiskreter Steuerungen mit PDEVSRCP Version 2.0 (WIP)
Birger Freymann, Artur Schmidt, Sven Pawletta, Sven Hartmann, Thorsten Pawletta
Birger Freymann
birger.freymann@hs-wismar.de
www.mb.hs-wismar.de/cea/
Agenda
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
2
Agenda
3
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
Motivation
• Nowadays control development is complextesting, maintenance, extension, …
• Ad-hoc implementations are often not possible
Design methodology:
Rapid Control Prototyping (RCP) approach by Abel (RWTH)
?
4
Agenda
5
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
Rapid Control Prototyping Approach
Rapid Control Prototyping Approach• Model based approach• Continuous control development • Avoidance of re-implementations ?
6
event based M&S approach !
usage of PDEVS for real time control
7
Agenda
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
Parallel Discrete EVent System Specification
8
PDEVS (Chow 1996) is an extension of DEVS (Zeigler 1976)
• Event-oriented modeling approach based on modular hierarchical model specification
• PDEVS - two system types:
atomic PDEVSdynamic behavior
coupled PDEVS (PDEVN)composition of atomic or coupled PDEVS
S
PDEVS Formalism
PDEVS & PDEVSRCP
9
𝑃𝐷𝐸𝑉𝑆𝑅𝐶𝑃 = 𝑋, 𝑌, 𝑆, 𝛿𝑖𝑛𝑡, 𝛿𝑒𝑥𝑡 , 𝛿𝑐𝑜𝑛, 𝜆, 𝑡𝑎
--------------------------------------------------------------------------
𝑃𝐷𝐸𝑉𝑆𝑅𝐶𝑃 = 𝑋, 𝑌, 𝑆, 𝛿𝑖𝑛𝑡, 𝛿𝑒𝑥𝑡 , 𝛿𝑐𝑜𝑛, 𝜆, 𝑡𝑎, 𝐴 (RG CEA)
PDEVS Formalism
𝑋 = 𝑋𝑚𝑜𝑑𝑒𝑙 ∪ 𝑋𝑐𝑙𝑜𝑐𝑘𝑋𝑚𝑜𝑑𝑒𝑙 = {(𝑝, 𝑣)|𝑝 ∈ 𝐼𝑃𝑜𝑟𝑡𝑠, 𝑣 ∈ 𝑋𝑝}𝑿𝒄𝒍𝒐𝒄𝒌 = {("𝒄𝒍𝒐𝒄𝒌", 𝒗)|𝒗 ∈ ℝ+}
𝐴 = 𝑎1, 𝑎2, … , 𝑎𝑛 set of executable activities𝑎𝑛 = (𝑎𝑖,[𝑡𝑖𝑚𝑖𝑛,𝑡𝑖𝑚𝑎𝑥])
𝜆 ∶ 𝑆 → 𝑌 × 𝐴 combined output and activity function
Agenda
10
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
Open Problems of PDEVSRCP 11
Dynamic PDEVS components with• ta(s) ∈ {0, ∞}• ta(s) ∈ ℝ+
ta(s) ∈ {0, ∞}
Open problems:(i) Specification of activities(ii) Specification of the Real Time Clock (RTC)(iii) How are activities specified within
atomic PDEVSRCP components ?(iv) Dynamic is limited (ta(s) ∈ {0, ∞})
𝑷𝑫𝑬𝑽𝑺𝑹𝑪𝑷 = 𝑿, 𝒀, 𝑺, 𝜹𝒊𝒏𝒕, 𝜹𝒆𝒙𝒕, 𝜹𝒄𝒐𝒏, 𝝀, 𝒕𝒂, 𝑨
atomic RTCPDEVSRCP
WCT?
?
x
𝑋𝑐𝑙𝑜𝑐𝑘 = {("𝑐𝑙𝑜𝑐𝑘", 𝑣)|𝑣 ∈ ℝ+}
re-implementation
12
Agenda
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
PDEVSRCP 2.0 Formalism13
activities
synchronous asynchronous
PDEVSRCP 2.0 Formalism
𝒂𝒊 = 𝑴,𝑿, 𝑺, 𝒀, 𝒇𝒆𝒙𝒕, 𝒇𝒊𝒏𝒕, 𝒇𝒐𝒖𝒕𝑴 Set of activity methods
𝑿 Set of input values
𝑺 Set of states
𝐘 Set of output values
𝒇𝒆𝒙𝒕 𝑓𝑒𝑥𝑡: 𝑆 × 𝑋 → 𝑆
𝒇𝒊𝒏𝒕 𝑓𝑖𝑛𝑡:𝑀 × 𝑆 → 𝑆
𝒇𝒐𝒖𝒕 𝑓𝑜𝑢𝑡: 𝑆 → Y
𝐏𝐃𝐄𝐕𝐒𝐑𝐂𝐏𝟐. 𝟎 = 𝑿, 𝒀, 𝑺𝑹𝑪𝑷, 𝜹𝒊𝒏𝒕𝜹𝒆𝒙𝒕, 𝜹𝒄𝒐𝒏, 𝝀𝑹𝑪𝑷, 𝒕𝒂
𝑆𝑅𝐶𝑃 = 𝑆 ∪ 𝑆𝐴𝑆𝐴 = 𝑎1, … , 𝒂𝒊, … , 𝑎𝑛 //set of activities
𝜆𝑅𝐶𝑃: 𝑆𝑅𝐶𝑃 → 𝑌 × 𝑆𝐴 // combined output and activity function
activity 𝒈𝒆𝒕𝑾𝑪𝑻 = 𝐌,𝑿, 𝑺, 𝒀, 𝒇𝒆𝒙𝒕, 𝒇𝒊𝒏𝒕, 𝒇𝒐𝒖𝒕𝑀 = 𝑟𝑒𝑎𝑑𝑆𝑦𝑠𝑡𝑒𝑚𝐶𝑙𝑜𝑐𝑘𝑋 = ∅ ⇒ 𝑓𝑒𝑥𝑡 not defined𝑆 = 𝑡𝑉𝑎𝑙𝑢𝑒 with s0 = 0𝑌 = 𝑡𝑤𝑐𝑡𝑓𝑖𝑛𝑡 𝑟𝑒𝑎𝑑𝑆𝑦𝑠𝑡𝑒𝑚𝐶𝑙𝑜𝑐𝑘, 𝑡𝑉𝑎𝑙𝑢𝑒
𝑡𝑉𝑎𝑙𝑢𝑒 ≔ 𝑟𝑒𝑎𝑑𝑆𝑦𝑠𝑡𝑒𝑚𝐶𝑙𝑜𝑐𝑘𝑓𝑜𝑢𝑡 𝑡𝑉𝑎𝑙𝑢𝑒
𝑡𝑊𝐶𝑇 ≔ 𝑡𝑉𝑎𝑙𝑢𝑒
14
𝑹𝑻𝑪 = 𝑿, 𝒀, 𝑺𝑹𝑪𝑷, 𝜹𝒊𝒏𝒕𝜹𝒆𝒙𝒕, 𝜹𝒄𝒐𝒏, 𝝀𝑹𝑪𝑷, 𝒕𝒂𝑋 = ∅, 𝑌 = ∅ ⇒ 𝛿𝑒𝑥𝑡 , 𝛿𝑐𝑜𝑛 not defined𝑆𝑅𝐶𝑃 = 𝑆 ∪ 𝑆𝐴; 𝑆 = 𝜎, 𝑡𝐿𝑎𝑠𝑡 , 𝑡𝑊𝐶𝑇 ; 𝑆𝐴 = 𝒈𝒆𝒕𝑾𝑪𝑻𝑠𝑅𝐶𝑃,0 = (0,0,0, 𝑔𝑒𝑡𝑊𝐶𝑇)𝑡𝑎 𝜎, 𝑡𝐿𝑎𝑠𝑡 , 𝑡𝑊𝐶𝑇 , 𝑔𝑒𝑡𝑊𝐶𝑇 := 𝜎𝜆𝑅𝐶𝑃(𝜎, 𝑡𝐿𝑎𝑠𝑡, 𝑡𝑊𝐶𝑇 , 𝑔𝑒𝑡𝑊𝐶𝑇)
𝒈𝒆𝒕𝑾𝑪𝑻. 𝒇𝒊𝒏𝒕(𝒈𝒆𝒕𝑾𝑪𝑻.𝑴,𝒈𝒆𝒕𝑾𝑪𝑻. 𝑺)𝛿𝑖𝑛𝑡 𝜎, 𝑡𝐿𝑎𝑠𝑡 , 𝑡𝑊𝐶𝑇 , 𝑔𝑒𝑡𝑊𝐶𝑇
𝒕𝑾𝑪𝑻 ≔ 𝒈𝒆𝒕𝑾𝑪𝑻. 𝒇𝒐𝒖𝒕 𝒈𝒆𝒕𝑾𝑪𝑻. 𝑺𝑖𝑓 𝑡𝑊𝐶𝑇 − 𝑡𝐿𝑎𝑠𝑡 < 0.1yes: 𝜎:= 0 //VT advance = 0no: 𝜎:= 0.1; 𝑡𝐿𝑎𝑠𝑡: = 𝑡𝑊𝐶𝑇 //VT advance = 0.1
atomic RTCPDEVSRCP 2.0
PDEVSRCP 2.0 Formalism15
Virtual TimeWal
l C
lock
Tim
eReal Time Clock (RTC)
16
Agenda
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
Application Example
SM PDEVS
CS PDEVSRCP
17
GEN GeneratorPROC ProcessorTRA TransducerINTF InterfaceRTC Real Time ClockSM Simulation ModelCS Control Software
input puffer robot
output puffer
Application Example
CS PDEVSRCP 2.0
SM PDEVS
CS PDEVSRCP
18
GEN GeneratorPROC ProcessorTRA TransducerINTF InterfaceRTC Real Time ClockSM Simulation ModelCS Control Software
19Application Example
virtual robot(s)
CS based on PDEVSRCP 2.0 (MATLAB/DEVS)
20
Agenda
1. Motivation
2. Rapid Control Prototyping Approach
3. PDEVS & PDEVSRCP Formalism
4. Open Problems of PDEVSRCP
5. PDEVSRCP 2.0 Formalism
6. Application Example
7. Summary and Outlook
21
Summary:
Analysis of first version of PDEVSRCP
Highlighting of open problems of PDEVSRCP
Definition and implementation of PDEVSRCP 2.0
Comparision of both PDEVSRCP variants
Outlook:
Classification of PDEVSRCP 2.0 in context of system therory
Usage of the new concept for robot control development
Summary and Outlook
[1] D. Abel, A. Bollig, Rapid Control Prototyping, Methoden und Anwendung, Springer, 2006. [2] G. Maletzki, Rapid Control Prototyping komplexer und flexibler Robotersteuerungen auf
Basis des SBC-Ansatzes. Diss., Uni. Rostock/HS Wismar, 2014. [3] B.P. Zeigler, H. Praehofer, T.G. Kim, Theory of modeling and simulation 2nd ed., Acad. Pr,
2000. [4] S.M. Cho, T.G. Kim, Real Time Simulation Framework for RT-DEVS Models, Trans. Soc.
Comput. Simul. Int. 18 (4) (2001) 203–215. [5] J.L. Risco-Martín, S. Mittal, J.C. Fabero, P. Malagón, J.L. Ayala, Real-time
Hardware/Software Co-design Using Devs-based Transparent M&S Framework, in: Proc. of the Summer Computer Conf., 45:1-45:8.
[6] T. Schwatinski, T. Pawletta, S. Pawletta, C. Kai-ser: Simulation-based development andoperation of controls on the basis of the DEVS for-malism. Proc. 7th EUROSIM Conf., Vol.2, Prag, Czech Rep., 2010, 8 pages.
[7] A.C.-H. Chow, Parallel DEVS: A Parallel, Hierarchical, Modular Modeling Formalism and Its Distributed Simulator, Trans. Soc. Comput. Simul. Int. 13 (2) (1996) 55–67.
[8] B. Freymann, T. Pawletta, S. Pawletta, Multi-Robotersteuerungen mit variablen Interaktionsprinzipien auf Basis des Simulation Based Control Frameworks und dem Dicrete Event System Specification Formalismus, Proc. ASIM STS/GMMS WS, ARGESIM Rep. No.50, 2015, 67–77.
[9] B. Freymann, T. Pawletta, T. Schwatinski, S. Pawletta, Modellbibliothek für die Interaktion von Robotern in der MATLAB/DEVS-Umgebung auf Basis des SBC-Frameworks, Proc. ASIM STS/GMMS WS, ARGESIM Rep. No.42, 2014, 199–208.
Thank you for your attention 22
23
Backup
coupled PDEVN
ActivityActivity
CLOSURE UNDER COUPLING
24
PDEVS
Activity
PDEVSRCP