- Cognitive Manufacturing System - June 21 st , 2012 Prof. Dr. - - PowerPoint PPT Presentation
a PLM 2012 a - Cognitive Manufacturing System - June 21 st , 2012 Prof. Dr. -Ing. HongSeok Park Laboratory for Production Engineering School of Mechanical and Automotive Engineering
a PLM 베스트 프랙티스 컨퍼런스 2012 a
Arriving Work-piece Loading machine Begin Machining
Machining
Finish Machining Unloading machine Next Machines
Disturbance
Tool-wear Tool-break
Current manufacturing
Parameter change Tool change Intervention of human
Adjusting parameter
request
Agent #1 Tool-wear Tool-break Agent #n
New manufacturing
Intelligent& Genetic behaviours Self- adaptive Cooperation Stop machine Experience Downtime
vReducing productivity vDecreasing the utilization
vMeasures depending on the experience of operators vSelf-adapting vReasoning ability in decision making vSelf-controlling ability
Stop machine Self-adapting system
Clutch housing
Clutch housing
vProcessing operations per product: 17 vMachines: 12
{ Downtime: 20-25% of total planned time
Recovery method
Current recovery method:
Stop the machining shop to repair and reset
Centralized control system § Rigid Control: Top-down problem solving § Low scalability § Low adaptability
Proposed method: Self-adaptive manufacturing system
Machine Agent 1 Machine Agent 2 Work-piece Agent Transporter Agent Machine 1 Machine 2 Work-piece Transporter …
Rescheduling type: Recovering time > 1 hour
Non-negotiation Negotiation Rescheduling 11.4% 47.7% 40.9%
Event Malfunction of machine (long recovering time)
Rescheduling
Agents
Event Tool-break
Agent
Negotiation Non-negotiation
Malfunction of machine (short recovering time) Event Tool-wear
Agent
Event
Disturbance
Disturbance Classification MES Disturbance Information
vData collection time: 2006.08.31-2009.08.18 vDisturbance numbers: 685
Disturbance class Type of disturbance Related to resources Machine breakdown Maintenance of machine Tool breakdown Tool wear Operator absenteeism Related to orders Unavailability of raw material Cancellation order Rework Arrival of a new job order Urgent job Delay in transport using material handling system Out sourcing Related to measurement of data Process time variation Variation of set-up times Change of priority Control software and Communication networks Malfunction
Non-Negotiation type: Recovering time < 30 mins Negotiation type: 30 mins < Recovering time < 1 hour
Disturbance
Rule- based Decision making
Conventional agent Percepts Actions Sensors Effectors Environment Agent technology Cognitive technology Perception Reasoning Actions Synthesis of agent and cognitive technologies
Perception (Beliefs) Interpretation Learning Decision Making (Desires) Action Communication Event Knowledge & Experience (Intentions) Reasoning Event recognition Information input New situation Update knowledge Familiar situation Plan Command
Cooperation
BDI Architecture Beliefs Grasping the information of the current states of an agent’s environment Desires All the possible states of tasks that agent could carry out Intentions The states of tasks that the agent has decided to work towards
Ability Machine tool Task1 Machine 1 Task 2 Machine 2 Task 3 Machine 3 Task 2 Machine 1 Task 2 Machine 3 Ability list Cognitive agent Ability Pheromone Task 1 Pheromone value 1 Task 2 Pheromone value 2 Information Node M1,T1 M2,T2 M3,T3 M1,T1 M3,T2,T3 Machine breakdown Route 2 Work-piece Product
Ants lay pheromone on the trail when they moves food back to nest Pheromones accumulate with multiple ants using same path Pheromones evaporate when no ant pass Ant travel rule: Each ant always try to chose the trail has higher pheromone concentration
From Natural to Manufacturing Systems
The machine with the shortest processing time for carrying out a specific
Product and Machining Shop Inspired Biology : Ant Colony Algorithm Cognitive Agent Elementary technologies to develop a SAMS Developing Architecture Of the machining shop based on Cognitive agents Model of the machining shop based on functional agents Information Model of SAMS Mechanism of SAMS for adapting to disturbance Algorithm for making decision of SAMS Implementing the Test-bed of SAMS Architecture of the Test-bed Mechanism of the implemented SAMS for adapting to disturbances Disturbance Analysis Disturbance Classification Finding measures against the corresponding disturbances
Model of SAMS Model of SAMS Test-bed Test-bed Strategies for overcoming disturbances Strategies for overcoming disturbances
Non-negotiation Negotiation
Diagnosis Disturbance Cognitive Agent Controller
Machine Agent #i Machine #i Machine #3 Perception Decision Making Control Communication
Machine Agent #1 Reasoning Signals Tasks Plan
Machine #1 Disturbance Work-piece
MES
Transporter
Machining system
Work-piece Agent Transporter Agent
Interpretation
Machine Agent #3 Machine #2 Machine Agent #2
vBehaviours policies § Rule-based § Reasoning mechanism vPheromone value
Knowledge
Machine #1
Machining Shop
Work-piece Transporter Machine #2 Machine #2
Negotiation Mean-ends reasoning Deciding on How to achieve this state
Previous state I, B Update state b:=see B:=update( B, b) Comparison t:= compare (B, D) MES D:=process(task) Inform normal state (t:=0) Disturbance type c:=Diagnosis (type) Type A: rescheduling Disturbance I:=filter(B,D,I) p:=plan(B, I, c) Rule-base Type B: Reactive behavior Execute (p) End Negotiation Select (agents) + Type C: Cooperative behavior without (p) Agent (selected) Execute (p) End MES without (agent)
Deliberation reasoning Deciding on what state to achieve Deciding on what agent to be selected
rescheduling
Hardware architecture of SAMS
Working sequence for implementing the SAMS
Software architecture of SAMS Developing the database of SAMS Non-negotiation mechanism of SAMS Negotiation mechanism of SAMS Evaluation of SAMS Demonstration of SAMS
PLC S7-300 Disturbance inputs
RS232 cable
light indicator
DI D0 CP341 module CP343-1 module Internet cable Internet card IP: 192.168.0.30
RFID Tag RFID Reader Work- piece Light
Read/Write Message for SEND/RECEIVE data
Flowchart SEND/RECEIVE between PLC and RFID
Agent #1
IP: 192.168.1.1 IP: 192.168.1.2 IP: 192.168.1.n
Agent #2 Agent #n Bridge in wireless network
IP: 192.168.1.3 Agent #3 Wireless Access point Access point Access point Access point
The communication bridge has function of the signal amplifier, and is a middleware node in the communication wireless network.
The communication mechanism between agents in the AMS
SQL KepserverEX OPC PLC1 Agent#1
Disturbance Disturbance information Work-piece information Process Information Task Knowledge
SQL KepserverEX OPC PLC2 Agent#2 SQL KepserverEX OPC PLC3 Agent#3 MES
Message Message Process Information Disturbance Resource Information
SQL
Wire Wireless
Machine1 RFID Machine2 RFID Machine3 RFID
Machine agent state
Information collection OPC items
A A
Disturbance information Disturbance classifying
Perception Decision
Plan Task Information
Cooperation information
A B A B A disturbance known B
disturbance unknown Suggested method Unknown Disturbances Process Information Machine Information
Non-Negotiation Negotiation
1
Input disturbance shown by turning on the red light (alarm) at PLC 1
6
Simentic S7
PLC #1 2 PLC 1 sends
the signal to agent
3 4 5 6
The system
disturbance shown by turning
at PLC 1 OPC protocol is used for communicating between PLC and agent In case of the disturbance belongs to the non-negotiation type, agent generates a new plan and sends the command to PLC
Collecting data
Agent diagnoses the disturbance type
Machine agent
PLC 1
MES
Registering network Input disturbance shown by turning on the red light (alarm) at PLC 1
1
PLC 1 sends the signal to agent Collecting data
2 3
Agent diagnoses the disturbance belonging to the negotiation type Agents establish the wireless network to server
4 5 Agent 1 Agent 2
(Selected agent) Agent 3
PLC 2
Agent negotiation An appropriate agent is selected for carring
failure machine
6 7
The system
disturbance shown by turning
at PLC 2
8
§ The cognitive agent technology and the biology inspired strategy are applied to the SAMS § Disturbances and corresponding management methods in the machining shop of a clutch housing are analyzed § Developing a SAMS to autonomously overcome these disturbances § Implementing a test-bed to demonstrate the functionalities of SAMS. § Implementing self-evolution mechanism for solving the new disturbance to be happened
§ This method could replace the traditional method that has been intervened by human operator § It has the functionalities of intelligent behaviour such as self-adapting, self-controlling, and reasoning ability in decision making § Increasing the productivity and reducing downtime in the product line.