Jamshidi December 16th Lecture

Slide1: 

Some Application Opportunities for Soft Computing Mo Jamshidi Electrical and Computer Engineering Department and Autonomous Control Engineering (ACE) Center University of New Mexico, Albuquerque Advisor, NASA HQ, US DOE HQ and US Air Force Research Lab. http://ace.unm.edu & http://vlab.unm.edu jamshidi@eece.unm.edu & moj@cybermesa.com December 16, 2003 – UC Berkeley

Slide2: 

An Opinion! Soft computing has great potentials in many current and future areas to help solve critical problems in today’s global environment. A combination of intelligent tools of SC, high-technology areas as well as common-sense approaches to problem solving will be a very working ingredient to help such areas as industry, space exploration, defense, security and economy.

Slide3: 

Some application opportunities of SC … Multi-agent systems (e.g. V-Lab® project) 2) Industrial energy efficiency 3) Process Control Systems (Petrochemical, chemical, liquid systems, etc.) 4) Diagnostic / Prognostics of hardware systems, with applications in space transport systems (Columbia Columbia Shuttle Disaster prevention measures), air transport (black boxes), national defense, etc. 5) Applications to “Return to Flight” and Space safety systems (NASA Langley located NESC – NASA Engineering and Safety Center (a 10-fields Center)

USA’s “Industries of the Future”(Most inefficient ones - IOFs): 

USA’s “Industries of the Future” (Most inefficient ones - IOFs) Steel Aluminum Glass Chemical Petroleum Metal Casting Mining -- UBC efforts Forest -- BC efforts Others – Food, Agriculture and Cement

USA’s “Industries of the Future”(IOFs): 

USA’s “Industries of the Future” (IOFs) For the past 6 months, four teams of scientists/engineers have been studying the future applications of 4 technologies : robotics, control, automation and information technology on energy efficiencies of IOFs.

Robotic Study Team: 

Robotic Study Team George Bekey, Prof. Emeritus, Univ. of Southern California Clarence de Silva, NSERC Prof., Univ. of British Columbia Jeanne Deutsch, CEO, ActiveMedia Robotics, Inc. Joseph Engelberger, CEO Emeritus, Unimation and HelpMate Robotics Mo Jamshidi, University of New Mexico, Chair DOE Robotics Team Part of the DOE- Energy Efficiency and REnewable (EERE) of Industrial Technologies Study Group: Industrial Control – Frank Doyle, University of California @ Santa Barbara Information Processing – Tunde Ogunnaike, University of Delaware 3. Automation – Mark Body & Bonnie Bennett, Adventium Labs.

BACKGROUND: 

BACKGROUND US Department of Energy’s Office of Energy Efficiency and Renewable Energy (EERE) of Industrial Technologies commissioned the above 4 aforementioned key technologies to assess their roles in the 7 of 8 energy-intensive IOFs : Aluminum, Chemicals, Forest Products, Glass, Metal Casting, Mining and Steel. Agriculture, Cement and Food Processing industries are of considerable interest to DOE, as is petroleum.

OBJECTIVE: 

OBJECTIVE … to identify robotics research funding by DOE/EERE office. … prioritize these opportunities by energy savings and energy efficiency.

“Strategic Plan” … US DOE ITP/EEREAugust 2003: 

“Strategic Plan” … US DOE ITP/EERE August 2003 Mission: “… to reduce our Nation’s reliance on foreign sources, reduce environmental impacts, increase the use of renewable energy, …” Vision:”… U.S. goods to be recognized for extraordinary quality and produced with minimal energy…” Goals: “… between 2002 and 2020, contribute to a 30% decrease in energy intensity … of the energy intensive industries of the future (a potential savings of 3.8 – 4.5 quads) ...”

ROBOTS and Energy Efficiency: 

ROBOTS and Energy Efficiency Some robotics effects on energy efficiency are indirect, i.e. Robots can solve an industrial problem which would make expenditure of energy unnecessary, e.g. mining robots, extreme temperature labor robots, AGVs, etc.. Robots can: Avoid production of defective products Reduce waste Reduce energy requirements for automation and remote operations Increase efficiency of existing processes via sensing and IT systems where demand and automated production modeling can be integrated, etc.

RECOMMENDATIONS: 

RECOMMENDATIONS Extreme-Temperature Robotic Systems = Applications – Furnaces, Boilers, Kilns, Dryers, etc. = Accumulated energy saving is 2.02 quads for 10 years, assuming 10% – 30% - 50% adaptation rule Major Opportunity Energy Efficient Robots = Redesign of all aspects of industrial and mobile robots for IOFs = Case - the estimated 60,000 hydraulic robots used in industries are very inefficient and their fundamental redesign can save accumulated 1.95x1012 BTUs (quads) by 2010. = Energy saving for electric robots is approximately 0.025 accumulated quads by 2010. Major Opportunity

RECOMMENDATIONS: 

RECOMMENDATIONS - Energy Efficiency Through Labor Robots = Robotic technologies present enormous opportunities for steel, mining, aluminum and forest industrial operations. = Labor robots eliminate the need for energy needs to heat, cool or condition workplace like mines, factory floors, etc. = Potential accumulated energy saving is 0.55 quads for 10 years, assuming same adaptation rule In all areas, a combination of IT, Soft Computing, sensor and control technologies on these robots will use of an integrated approach to energy efficiency.

A SC example in forest industry: 

A SC example in forest industry FOREST PRODUCTS INDUSTRY (de Silva) The industries of forestry, lumber, wood products, pulp and paper, and fuel wood are the main constituents of the Forest Products Industry. This industry makes a major contribution to the Nation's economy and employment base. It employs close to 2 million workers in the United States, generates annual sales of about $250 billion (about 10% of which is through export), and represents nearly 10% of the US manufacturing output. About $100 billion of the annual sales is from lumber and associated wood products, about $60 billion is from pulp and paper, and about $40 billion is furniture and similar finished products. The number of pulp and paper mills in the United States is less than 1000, while there are about fifty thousand sawmills, secondary mills, and wood product plants.

A SC example in forest industry: 

A SC example in forest industry The energy needs and self-generation in some important industries.

A SC example in forest industry: 

A SC example in forest industry Heater Fan

A SC example in forest industry: 

A SC example in forest industry Block diagram of the Fuzzy-PID Dry Kiln control system

A SC example in forest industry: 

A SC example in forest industry Energy savings through innovative fuzzy-PID control. For 3% Moisture Removal Heating Duty Cycle goes from 50% (PID) to 15% (Fuzzy-PID) For 5% Moisture Removal Heating Duty Cycle goes from 90% (PID) to 30% (Fuzzy-PID)

SC Applications in ROBOTICS Needs in 10 US Industries : 

SC Applications in ROBOTICS Needs in 10 US Industries Need Index: 0 - Lowest need, 10 – Highest need

 SC -Based Data Mining Algorithms for Prognostic Studies of the Hardware System1: 

SC -Based Data Mining Algorithms for Prognostic Studies of the Hardware System1 Approaches to Diagnostics and Prognostics Data Driven Methods Analytical Methods Knowledge Based Methods ____________________________________ 1 – This work is being done in collaboration with IIS Corporation and its CEO Dr. H. Berenji and Prof. R. Langari, Texas A&M University, Work is being supported by both USAF and MDA.

D-P Driven Hardware System: 

D-P Driven Hardware System PLANT (Shuttle Subsystem) Prognostic Subsystem Outputs Warning ! Inputs Resolution Commands Mined Data Data Mining Pattern Recognition Prognostic Subsystem SC Tools

Data Driven Methods: 

Data Driven Methods Feature extraction: Partial Least Square (PLS) Fisher Discriminant Analysis Canonical Variate Analysis Principal Component Analysis Current work will only focus on PCA and its non-linear relative (NLPCA).

Integrated Method for Fault Diagnostics and Prognostics (IFDP): 

Integrated Method for Fault Diagnostics and Prognostics (IFDP) Based on NLPCA for dimensionality reduction Society of experts (E-AANN, KSOM, RBFC) Gated Experts All developed in Matlab with Simulink for model simulations (Future plans will be to do it in V-Lab® as well)

Extended Auto-Associative Neural Networks (E-AANN): 

Extended Auto-Associative Neural Networks (E-AANN)

Kohonen Self-Organizing Maps (KSOM): 

Kohonen Self-Organizing Maps (KSOM) KSOM defines a mapping from the input data space n onto a regular two-dimensional array of nodes. In many plants, a KSOM input is a vector combining both inputs and outputs of a certain system component. Every node i is defined by a prototype vector mi  n. Input vector x  n is compared with every mi and the best match mb is selected.

Kohonen Self-Organizing Maps (KSOM): 

Kohonen Self-Organizing Maps (KSOM) Three-dimensional input data in which each sample vector x consists of the RGB (red-green-blue) values of a color vector.

Chiller Model at Texas A&M University: 

Chiller Model at Texas A&M University

Hardware System 1 – CNC Machine: 

Hardware System 1 – CNC Machine MAIN POWER TRANSFORMER SPINDLE DRIVE AXIS AMPLIFIER DC POWER SUPPLY CNC CONTROLS XYZ CURRENT FEEDBACK XYZ ENCODER FEEDBACK XYZ SPEED COMMAND ENCODER FEEDBACK Y Z X

Hardware System 1 – CNC Machine: 

Hardware System 1 – CNC Machine CPU VIDEO COMPUTER X Y Z S INTERFACE

Laser Pointing System at UNM: 

Laser Pointing System at UNM L A S E R Lab View Fuzzy Controller Algorithm ADC DAC DAC X/Y motors Detector Quadrant Mirror Filter L A S E R

Laser Pointing System at UNM, Cont’d.: 

Laser Pointing System at UNM, Cont’d.

An Industrial Laser Prognostic System : 

An Industrial Laser Prognostic System Knowledge Base (NOP Senior Engineers) Laser System Subsystem PCA Data Reduction Expert System Original Data RBFC Inputs Reduced Dominant Data KSOM Relevant Data E-AANN Inputs GE-NN System Diagnostic – Prognostic System Outputs Fuzz IEEE Budapest

D-P Using SC Tools - summary: 

D-P Using SC Tools - summary Due to the huge number of sensors on large industrial or national systems SC approaches for fault diagnostics and prognostics must be capable of intelligent data reduction in such a way that no important data is lost and all the crucial data be used for smart (SC based) prognosis with minimum false alarms. It is expected that a library of these strong methods which is under development at IIS Corp and University of New Mexico will significantly benefit many hardware systems health, including military, industrial and other systems. A new SBIR proposal by IIS and our team has just been approved.

Role of SC in Complex Process Control Systems … : 

Role of SC in Complex Process Control Systems … Large-scale systems are associated with three concepts: 1. Decomposition 2. Centrality 3. Complexity SC can be utilized to make analysis and design of these systems possible.

Slide34: 

Process Industry is among those which have shun away from SC approaches for design.

Autonomous Large-Scale Systems - Hierarchical : 

Autonomous Large-Scale Systems - Hierarchical Fuzzy Logic Rules Coordinator Subsystem 1 … a1 an {x1,u1} {xn,un} interaction factor state, control Neuro-Fuzzy Controller 1 Subsystem n Neuro-Fuzzy Controller n

Autonomous Large-Scale Decentralized Systems : 

Autonomous Large-Scale Decentralized Systems LARGE-SCALE SYSTEM Fuzzy Controller 1 Fuzzy Controller n u1 un y1 yn . . . Output Input

Autonomous Large-Scale Systems – Adaptive Decentralized : 

Autonomous Large-Scale Systems – Adaptive Decentralized LARGE-SCALE SYSTEM PID Controller 1 u1 un y1 yn . . . Output Input y1d Fuzzy Tuner 1 PID Controller n Fuzzy Tuner n ynd

Autonomous Control of a Water Treatment Plant: 

Autonomous Control of a Water Treatment Plant

Another example an autonomous system: 

Another example an autonomous system A system, a device or a vehicle is said to be operating under autonomous behavior if it is in an state of “autonomy”, i.e. can guide itself for any future action or decision for a period of time. An autonomous system is under the guidance of autonomous control. Best example is the Pathfinder which landed on Mars on July 4, 1997. Operated semi- autonomously for 92 days.

LISA - Advanced Avionics Systems for Dependable Computing in Future Space Exploration - Astrophysics: 

LISA - Advanced Avionics Systems for Dependable Computing in Future Space Exploration - Astrophysics Laser Interferometry Space Antenna (LISA)

Slide42: 

Mars 09 Smart Lander/Rover Mission HAZARD DETECTION/AVOIDANCE

Flyby Scenario for Distributed Satellites: 

Flyby Scenario for Distributed Satellites Scenario A Work to appear in Comp in IE Journal, 2004.

Fuzzy Image Enhancer: 

Fuzzy Image Enhancer Initial Image Fuzzy Expert System Enhanced Image

Slide45: 

Fuzzy Expert System Image Enhancer

Slide46: 

Image Enhancement System

ANALOG IMAGES - Human EXPERT: 

ANALOG IMAGES - Human EXPERT Exposure system controlled by a human expert

Automated process of enhancing an analog image: 

Automated process of enhancing an analog image

Software Flow Chart: 

Software Flow Chart

Analog Photo – Before($ 1+ M System at the Wal-Mart): 

Analog Photo – Before ($ 1+ M System at the Wal-Mart)

Analog Photo – After ($ 6.5 K ACE System): 

Analog Photo – After ($ 6.5 K ACE System)

Slide52: 

You are cordially invited to attend: WAC 2004 (Seville, Spain - www.wacong.com) June 28 – July 1, 2004 2) IEEE SMC 2004 (Hague, Netherlands) October, 10-13, 2004 (http://www.ieeesmc2004.tudelft.nl) and yes one more …

Slide53: 

and yes …. IEEE SMC 2005 ( Big Island, Hawaii ) October, 2005

Slide54: 

Thank You!