Modular Logic Controllers for Machining Systems: Formal Representation and Analysis using Petri Nets

Size: px

Start display at page:

Download "Modular Logic Controllers for Machining Systems: Formal Representation and Analysis using Petri Nets"

Garry Pope
5 years ago
Views:

1 Modular Logic Controllers for Machining Systems: Formal Representation and Analysis using Petri Nets Dawn Tilbury Mechanical Engineering and Applied Mechanics University of Michigan

2 Acknowledgments Joint work with Euisu Park and Pramod Khargonekar of University of Michigan Funding provided by NSF/ERC for Reconfigurable Machining Systems at UM Industrial partners from Lamb Technicon for motivation, examples, and feedback

3 Outline Motivation: Logic control problem for high-volume machining systems Background Automatic cycle logic control Multi-mode logic control Implementation and future work

4 High-volume machining systems Cycle time determined by projected demand 500,000 parts/year one part every 20 sec. One operation per station Dedicated transfer machine: Fixed material handling, part flow Highly accurate and repeatable operations stations, buffer at beginning/end 5,000-10,000 I/O points (sensors/actuators) transfer machines per transfer line

5 Control of machining systems Continuous variable control (servo, CNC) Positioning, feed for metal removal, fixturing, and transfer operations: mainly SISO loops Discrete-variable (Logic) control (PLC) Sequencing of operations Coordination and synchronization of stations User interface (pushbuttons & display) Fault diagnosis and fault handling Safety: gates and interlocks Machine services: coolant, lubrication

6 Logic control problems Control software is 50% of machining system development time and cost Often written in low-level language (ladder) Specification for automatic cycle: timing bar chart, but only requires 10% of control code No specification for alternate control modes, error handling, diagnostics No formal verification before implementation Long testing and debugging cycles result

7 Industry trends Faster time to market for new products Must reduce system development time Shorter product lifecycles Need to change manufacturing systems to produce new products Increased quality Integrate new technologies into existing manufacturing systems

8 Reconfigurable machining systems Modular: easy assembly, potential reuse Upgradable: new technology integration Convertible: quick changeover to new part

9 Logic control challenges in RMS Modular structure Same granularity as machine modules Ease of configuration/reconfiguration Mathematical basis Enable formal verification of correctness High-level abstraction for ease of understanding control system functionality Industrial implementation Realistic complexity (thousands of events) Target system: PLC or PC code generation

10 Outline Motivation: Logic control problem for high-volume machining systems Background Transfer lines and Petri nets Automatic cycle logic control Multi-mode logic control Implementation and future work

11 Transfer line example Modular, standardized machining systems Each station self-contained with controls Coordination through transfer bar

12 Control specification Timing bar chart specifies auto mode Developed by machine designers Operation sequence Causal dependencies Home operation Mechanical stability Hydraulics Cycle time Transfer Bar Cradle Clamp 1 Mill 1 Rotating Table Raise Transfer (1st Lift) Raise Transfer (2nd Lift) Raise Transfer (3rd Lift) Raise Transfer (4th Lift) Advance Transfer Lower Transfer (1st Lower) Lower Transfer (2nd Lower) Lower Transfer (3rd Lower) Lower Transfer (4th Lower) Return Transfer Advance Cradle (Machining Position) 2.9 Return Cradle (Transfer Position) 3.8 Advance Clamp Read Part Seated Air Checks Return Clamp Rapid Advance Positioning Slide 2.1 Decel OPERATION GPM SEC Servo Servo Feed Main Slide Servo Rapid Return Positioning Slide 1.3 Reset Main Slide Servo Lower Rotate Advance Grippers Raise Rotate Return Rails Advance Rotate 270 Deg. Return Grippers Return Rotate 270 Deg. Advance Rails Total; Cycle Time = 22.2 second

13 Logic control block diagram Inputs: sensors and operator commands Outputs: servo or hydraulic devices

14 Formal representation of control Petri net: Bipartite graph Places (states) model operations: active or idle Token (dot) marks active state Transitions model events: from sensors or operator commands

15 Advantages of Petri net models Large base of existing theory Verification of key properties of control Hierarchical structures for complexity management Implementation in PLC SFC = IEC standard programming language One-one translation from Petri net to SFC

16 Petri net model capabilities p1 t1 t2 p5 p6 p7 p8 p9 p2 p3 p4 p10 Causal dependency p1 occurs before p2 Conflict Only one of p3, p4 can occur Transition conditions t1, t2 must be mutually exclusive Concurrency p5, p6 active simultaneously Synchronization p10 cannot begin until p7,p8,p9 have finished

17 Petri net properties for logic control Live: every transition can eventually occur No deadlocks All operations and events can happen Safe: No more than one token per place Ongoing operations are not requested Boolean state representation (active, inactive) Reversible: Initial state always reachable Guarantees cyclic behavior of system

18 Supervisory control problem Petri net control models desired closedloop behavior of system Control actions generated based on desired state Control Actions Machining System Sensors Control Generator Desired States Petri Net Control Operator Commands

19 Outline Motivation: Logic control problem for high-volume machining systems Background Automatic cycle logic control Multi-mode logic control Implementation and future work

20 Auto mode Normal operation cycle Unclamp parts Raise transfer Advance transfer Lower transfer Clamp parts Cycle machining stations Return transfer Transfer Bar Advance Clamp Return Clamp Rapid Advance Positioning Slide 2.1 Overlapping to minimize cycle time Fault diagnosis and fault stop Cradle Clamp 1 Mill 1 Rotating Table Raise Transfer (1st Lift) Raise Transfer (2nd Lift) Raise Transfer (3rd Lift) Raise Transfer (4th Lift) Advance Transfer Lower Transfer (1st Lower) Lower Transfer (2nd Lower) Lower Transfer (3rd Lower) Lower Transfer (4th Lower) Return Transfer Read Part Seated Air Checks Decel OPERATION GPM SEC Advance Cradle (Machining Position) 2.9 Return Cradle (Transfer Position) Servo Servo Feed Main Slide Servo 9.7 Rapid Return Positioning Slide Reset Main Slide Servo 9.0 Lower Rotate Advance Grippers Raise Rotate Return Rails Advance Rotate 270 Deg. Return Grippers Return Rotate 270 Deg. Advance Rails Total; Cycle Time = 22.2 second

21 Timing bar chart -> Petri net Each mechanical module gives one directed circuit Add wait states before synchronized transitions to model operation termination Mill 1 Rapid Advance Positioning Slide Decel Feed Main Slide Rapid Return Positioning Slide Reset Main Slide 20mm Rapid Advance 10mm Decel. Mill Head Reset Main Slide 553mm Feed Engine Block 2.1 Servo 1.3 Servo mm Rapid Return W Wait to Start Synchronized Transition Rapid Advance 20mm Decel. 10mm Feed 553mm Rapid Return 30mm Reset Main Slide 553mm

22 Hierarchical representation Reduce complexity of Petri net by combining sets of places/transitions Preserves liveness, safeness, reversibility

23 Combining control modules Merge synchronized transitions Synchronized transitions retain physical meaning p4 Add wait states (W) p1 p5 t1 t3 before p2 synchronized p6 p3 transitions to t2 t4 model operation p7 p8 completion p1 p2 p3 p4 p5 t1 W p6 t2 W p7 W W p8 t3 t1 & t3 t4 t2 & t4

24 Transfer Bar Cradle Logic control for automatic cycle Directly generated from timing bar chart Guaranteed to be live, safe, reversible OPERATION GPM SEC Raise Transfer (1st Lift) Raise Transfer (2nd Lift) Raise Transfer (3rd Lift) Raise Transfer (4th Lift) Advance Transfer Lower Transfer (1st Lower) Lower Transfer (2nd Lower) Lower Transfer (3rd Lower) Lower Transfer (4th Lower) Return Transfer Advance Cradle (Machining Position) 2.9 Return Cradle (Transfer Position) Servo Servo Total; Cycle Time = 22.2 second Transfer Bar 1 2 Cradle Clamp Mill 1 Mill 2 Clamp 1 Advance Clamp Read Part Seated Air Checks Return Clamp Mill 1 Rapid Advance Positionong Slide 2.1 Decel Feed Main Slide Servo Rapid Return Positioning Slide 1.3 Reset Main Slide Servo Rotating Table Lower Rotate Advance Grippers Raise Rotate Return Rails Advance Rotate 270 Deg. Return Grippers Return Rotate 270 Deg. Advance Rails

25 Reconfiguration of logic control Add a new mill to transfer line New logic for mill & clamp Modify transfer bar logic to synchronize with clamp Shaded logic unchanged Transfer Rotating Bar Cradle Clamp 1Mill 1 Table Clamp 4 Mill 4

26 Modular structure of control Transfer bar synchronizes all modules Limited communication Ease of modification Cradle & Clamp for Mill 1&2 Mill 1 Mill 2 Rotating Table Transfer Bar Clamp for Mill 3 Mill 3 Clamp for Mill 4 Mill 4

27 Outline Motivation: Logic control problem for high-volume machining systems Background Automatic cycle logic control Multi-mode logic control What happens when things go wrong? Implementation and future work

28 Operator interaction with control Manual mode Single station fine operation control Hand mode Normal operation cycle without overlapping Reverse sequences possible Unclamp All Stations Raise Transfer Advance Transfer Lower Transfer Cycle All Machining Stations Return Transfer Cycle Rotating Table Station Clamp All Stations

29 Three states for each operation Operating: Activate actuator Completed stop: Deactivate actuator Set internal variable to 1 Incomplete stop: Deactivate actuator Announce fault IC Events Sensors trigger transitions between states Internal variable used for coordination with other stations O C Events

30 Internal variables to synchronize Synchronization conditions depend on control mode Decouple each control module O A1 B1 O O A1 B1 O C C C B1 A1 C O A2 B2 O O A2 B2 O C C C C

31 Reversible operation module Normal and fault recovery operations Fault recovery: Restart from fault position Return to initial position of operation, then restart Faults may occur in fault recovery operation Transitions triggered by operator commands advance (a) and return (r) in manual mode a r r a IC IC a r a O C O C Normal Operation sn and (r or a) r Fault-recovery Operation sf and (r or a)

32 Irreversible operation module Metal-removal operations (i.e. milling) Only contains normal operation block Fault recovery: Restart from fault position Return to home position of machining station and restart Transitions triggered by operator commands advance (a) and return (r) in manual mode a IC a home position O C Normal Operation sn and a

33 Combining control modes Same set of states for all modes Transition conditions depend on active control mode Superposition preserves liveness, safeness, reversibility properties p 1 t1 p 2 Auto p 1 t2 p 2 Hand p 1 t3 p 2 Manual t1 t2 p 1 p 2 t3 t p 1 p 2 t = (t1^auto) (t2^hand) (t3^manual) ^ ^

34 Algorithm to build logic controller 1. Assign module for each operation 2. Normal operation cycle ordered as in timing bar chart Mill example: Advance, Feed, Return IC O IC O IC O C C C IC O IC O C C

35 Algorithm to build logic control 3. Reverse sequences for repeatable steps in hand mode 4. Fault recovery sequences for irreversible operations Theorem: Station control module constructed according to algorithm is live, safe, and reversible Petri net.

36 Mode decision control logic Mode chosen by operator from input panel Mode decision control logic is live, safe, (All) reversible Main Console Station- Auto Station- Manual Master/Start Master/Stop Manual E-Stop System Stop (Home Position) (Any) Station- Manual E-Stop Auto Hand E-Stop Standby Master/Start Auto Auto Any Fault Each Station Hand Master/Stop Station- Auto Hand Auto Station- Manual Any Fault

37 Modular logic controller Operation causality condition to ensure well-ordered set of operations Station logic controllers with mode decision control logic Theorem: Resulting controller is guaranteed to be live, safe, reversible

38 Outline Motivation: Logic control problem for high-volume machining systems Background Automatic cycle logic control Multi-mode logic control Implementation and future work To the factory floor

39 Implementation in PLC Sequential function charts (SFC) IEC standard language Based on Petri nets One-one translation Simple Place Initial Place Simple Transition Synchronized Transition Marked Graph Macro Place SFC/Grafcet Simple Step Initial Step Simple Transition Synchronized Transition Macro Step

40 Industrial implementation US Patent applied for, 1998 Current cooperation with Lamb Technicon on Cummins Engine project Evaluating implementation needs Machine services Safety and gate interlocks Reusability PLC platform dependence User interface Commercialization potential

41 Future of reconfigurable control Unified framework for continuous and discrete control for machining systems Modular structures for reconfigurability Mathematical basis for verification Integrated diagnostics Automatic fault detection and recovery Software tools for control design/analysis Interface with mechanical design software Automatic control code generation

Technological Training Programs

Technological Training Programs On behalf of Noaman Engineering, I would like to introduce you to our training courses. All of our courses cover Theoretical, Practical, and software implementation and