Valuasi dan Komersialisasi Teknologi (2009)

Yandra Arkeman

Lien Herlina

Aji Hermawan

1

Perkuliahan:

1 jam kuliah (Jum’at) 3 jam responsi (Rabu)

Asisten : Nisa Zahra, Banun

2

Topik yang akan dipelajari :

Valuasi teknologi (hasil inovasi) Komersialisasi teknologi

Termasuk di dalamnya: Inovasi Teknologi Technopreneurship HAKI (Hak Atas Kekayaan Intelektual)

3

4

INNOVATION CULTURE IN JAPAN IN THE AREA OF INFORMATION AND COMPUTER TECHNOLOGY

A special gift from JAPAN for students of TIN from

YANDRA

5

INTRODUCTION

The key of success of Japan and other developed countries INNOVATION

Everyday scientists and engineers in Japan think about new products, new process, new management techniques and new methods for improving their quality of life

This innovation culture should also be adopted by students of Department of Agroindustrial Technology

6

Example of Recent Innovation in Japan High speed train (Shinkansen) :

250 km/hr 500 km/hr Maglev (Magnetic levitation systems)

Aeroplane jet engine Roll Royce USA Honda Japan (not only motor bike and car)

Robotics and Aerospace technology, etc IMTS (Intelligent Mode Transportation

Systems) driverless car

7

Some Photographs in Japan

8

9

INNOVATION IN AGROINDUSTRY Product Innovation

New products, new materials Examples: bioplastics, bio-diesel, bio-lubricants, leather

products, etc Process Innovation

New process, new technology for producing products Examples: Using new technology for producing bioplastics

from sweet-potato, etc System Innovation

New management techniques Examples: Supply chain management, lean and agile

agroindustrial systems, zero-waste management

10

Methods Innovation New methodology, new algorithms, new

information technology techniques Examples: artificial intelligence, meta-heuristics,

genetic algorithms, multiobjective optimization, etc

FOCUS OF MY POSTDOCTORAL RESEARCH IN JAPAN

(Details of my research results will be discussed in the following slides)

11

Research Topic : Innovation in Computer and Information

Technology (Genetic Algorithms) for Agroindustry Researchers :

Dr. Yandra (TIN-IPB, Indonesia) Prof. Hiroyuki Tamura (Kansai University, Japan)

Host Institution : Department of Electrical Engineering and

Computer Science, Kansai University

12

13

14

15

16

17

RESEARCH BACKGROUND

Many real-world problems in engineering design, agroindustry, decision making and information technology involve simultaneous optimization of multiple objectives

The principle of multi-objective optimization is different from that in a single objective optimization

Single objective optimization the goal is to find the best solution ( MAX or MIN)

Multi-objective optimization: No single optimal solution (due to conflicting objectives) Compromise solutions or Pareto-optimal solutions Then, choosing the most preferred solution for implementation

18

SINGLE OBJECTIVE OPTIMIZATION

x

Max

F (x).

19

Objective 1

Ob

ject

ive

2nadir solution

utopian solution

feasibleregion

infeasibleregion

Pareto-optimal solutions

MULTIOBJECTIVE OPTIMIZATION(Minimizing Obj.1 and Minimizing Obj.2)

Dominated solutions

20

The aim of this research is: to develop a new computer-based methodology for multi-

objective optimization to apply this new methodology in agroindustry and other areas

The development of this methodology consists of two steps: Developing a new multi-objective genetic algorithm for finding a

well diverse Pareto-optimal solutions Developing an expert system for selecting the most preferred

solution based on higher-level information

21

Genetic Algorithm

Expert System

Pareto-optimum solutions

Final solution

Intelligent DSS

Multi-objectiveproblems

Z1

Z1

Z2

Z2

22

Main Program

Initialize Generation ReportInit-Data

Init-Pop

Init-Report

Mutation

Crossover

Selection

Old-PopNew-Pop

New-Pop2

New-Pop3

NDS+CD

ADM

New-Pop3

Pareto-optimum solutions

23

One of the methods to process higher level information for selecting the most preferred solution is an expert system (ES)

The basic ideas of an ES: Expertise is transferred from a human into a computer

stored in a knowledge-base The computer can make inferences and arrive at

specific conclusion Then computer gives advices and explains, if

necessary, the logic behind the advice

EXPERT SYSTEMS

24

APPLICATION FOR AGROINDUSTRY

Agroindustrial Supply Chain Management (Agro-SCM) :

The management of the entire set of production, transformation/processing, distribution and marketing activities

in agroindustry by which a consumer is supplied with a

desired product Agro-SCM is more complicated than manufacturing SCM

agricultural products are perishable

25

..….

..…

…..

Agroindustry 1

Agroindustry j

Farming 1

Farming i

Customer 1

Customer k

Description of Agro-SCM Model

26

ICIFICIPXCAYCB i

I

iij

J

jjij

I

i

J

jijjk

J

j

K

kjk

TSCCMin

111 11 1

:

kk

J

jjk DY

1

jj

K

kjk CapY

1

ii

J

jij SX

1

jj

K

kjk

I

iij IYX

11

Subject to:

Mathematical Model for Agroindustry SCM

IPIFIPIPXPAYPB i

I

iij

J

jjij

I

i

J

jijjk

J

j

K

kjk

ENDPMin

111 11 1

:

27

Optimization objectives : First Objective:

Minimizing Total Supply Chain Cost (TSCC) that consists of transportation and inventory costs

Second Objective : Minimizing Expected Number of Deteriorated Product

(ENDP) Maximizing quality Essential for agroindustry

As both objectives are conflicting : No single optimum solution Pareto-optimum or non-dominated solutions or trade-off

solutions

28

Optimization variables : The amount of raw materials to be transported from

farming to agroindustry (Xij) The amount of products to be transported from

agroindustry to consumer (Yjk)

Inventories at agroindustry (Ij) and farming (Ii) Subject to some constraints, such as:

Demand constraint Supply constraint Inventory constraint Production capacity constraint

29

A case study 2-farming, 2-factory and 2-customer (2 x 2 x 2 SCM-problem)

Compare the performance of h-NSGA-II and original NSGA-II

30

12 9 4 23

YA1YA2 XPA XPB

Chromosome Structure for 2x2x2 Agro-SCM

31

12 9 4 23

8 17 41 15

12 9 41 15

8 17 4 23

P1

P2

C1

C2

Crossover

crossover point

32

12 9 41 15 C1

12 11 41 15 C1’

*

mutation point

Mutation

33

Multi-objective genetic algorithms were then executed with the following parameters: Crossover Probability (Pc) 0.9, 0.8, 0.7

Mutation Probability (Pm) 0.05, 0.01, 0 Population Size = 20 Number of replication = 5

Output of h-NSGA-II and original NSGA-II are presented in the following figures

34

14

14.5

15

15.5

16

16.5

800 850 900 950 1000 1050 1100 1150

TSCC

EN

DP

Pareto-optimum solutions of heterogeneous NSGA-II

35

14.2

14.4

14.6

14.8

15

15.2

15.4

15.6

15.8

16

800 850 900 950 1000 1050 1100 1150

TSCC

EN

DP

Pareto-optimum solutions of original NSGA-II

36

The above figures show that: Pareto-optimum solutions could be found by h-NSGA-II in a

reasonable number of generation (i.e. 50) The solutions’ diversity of heterogeneous GA is better than its

original version There are 20 solutions produced by h-NSGA-II compare to 14

solutions produced by NSGA-II

These Pareto-optimum solutions were then fed to an expert system for selecting the most preferred solution

37

RULES

Recommendations(A, B, …,T)

QS (H,M,L)

FB (H,M,L)

CI (E,G,F,P)

TP (A,B,C)

Dependence diagram of the Expert System

QS = Customer and social concern about product quality and safety

CI = Company’s cash inflow

TP = Type of product being handled

FB = Possibility to get additional funding or loans from Bank

38

The input for expert system are: QS High CI Good TP type C FB Medium

Rule 064 was fired Choose SCM D-20: YA1 = 20

YA2 = 30

XPA = 38

XPB = 50 TSCC = 1048 m.u and ENDP = 14.1 units

39

APPLICATION OPPORTUNITIES IN OTHER AREAS At this stage we have developed an intelligent and

integrated methodology for solving multi-objective problems

One application example Agro-SCM In principle this new methodology can be used in

many areas of research with minor adjustments This section :

Review recent developments of multi-objective optimization in the areas of engineering design and information technology

Discuss the application prospects of the methodology developed in this research in those particular areas

40

An example Design of Turbojet Engines (2005) Used modified version of NSGA-II for multi-objective

optimization of thermodynamic cycle of ideal turbojet engines

The multiple and conflicting thermodynamic objectives used in this research are: Specific thrust Thrust specific fuel consumption Propulsive efficiency Thermal efficiency

ENGINEERING DESIGN

41

This research can be enhanced by the introduction of an expert system for choosing the most preferred solution

Without this ES the task of selecting final solution for implementation becomes very difficult and complicated

42

Also known as grid computing Seeks to unify geographically dispersed systems to

create one large and powerful system An example Application of GA for solving Task

Assignment Problem (TSAP) in grid computing (2006) Used HNNGA to solve tasks assignment of programs to

a number of processors to minimize a cost function This research only considers one objective

DISTRIBUTED COMPUTING SYSTEMS

43

In fact, the objective of TSAP can be many, such as: Minimization of completion time of the entire programs (make-

span) Minimization of communication time among tasks Minimization of processor’s load, etc

These multiple (and conflicting) objectives can be solved using the intelligent and integrated methodology proposed in our research

44

One important area in today’s information technology is intelligent data mining systems

An example Application of multi-objective genetic algorithms for data mining (2004)

Used NSGA-II for optimizing rule extraction process Multiple objectives:

Maximizing confidence of the rules Maximizing coverage of the rules

In the future this research can be improved by using new techniques such as h-NSGA-II as well as expert systems

DATA MINING

45

RESEARCH FOR THE FUTURE

We have presented the development of h-NSGA-II and expert system for solving multiobjective problems

The usefulness of this methodology has been shown using an Agro-SCM case study

We have published 6 conference papers and submitted 1 journal paper. Another journal paper is in preparation now. Our computer programs are being copyrighted.

This research concludes that this new methodology is robust and reliable it can be used in many areas with some modifications

Further research we have formulized “Research Framework” for the future as presented in the next slide

46

Strategic Research Framework

Single ObjectiveGenetic Algorithms

Multi ObjectiveGenetic Algorithms

Development ofNew Methods

Real-worldApplications

General purpose GA software

Hybrid with the other AI tools

Chromosome representation

Genetic operators

Initial population formation

Population diversity, convergence

Supply chain management

Scheduling, System Design

Grid computing, Data mining

Other areas in Mfg, Agroindustry, IT

GA-brain ©TM

Neural Networks

Fuzzy Logic

Expert Systems

Simulation

Bioinformatics

Parallel Meta-heuristics

47

Institutions and Counterparts

Institutions: TIN-Fateta-IPB, Bogor, Indonesia Kansai University, Osaka, Japan

Prospective Counterparts: Centre for Advanced Manufacturing Research

(CAMR), University of South Australia Prof. Lee Luong, Dr. Sev Nagaligam

Institute of Intelligent Information and Communication Technology (IICT), Konan University, Japan Prof. Wuyi Yue

48

National Engineering Research Centre for Information Technology in Agriculture (NERCITA), Beijing Dr.Xuzhang Xue

University Putra Malaysia (UPM) Prof. Amin Mohd. Soom

Nanyang Technological University (NTU), Singapore Prof. Khoo Li Peng

Funding : JSPS Japan Society for Promotion of Science Other competitive sources (National and

International)

49

The ultimate goals (in the long term) : Truly intelligent machines Computers with their own

minds True electronic brains

for improving the qualityof human-life

50

Research Topics for Students (S1,S2,S3) of TIN/TIP Innovation Culture Application of Single and Multiple Objective Genetic

Algorithms for Agroindustrial Systems Design (S1,S2,S3) Scheduling Project Management Plant Lay-out and Design Production Planning, etc

Optimization of Agroindustrial Supply Chain Networks Using Intelligent Systems (S1,S2,S3) All commodities of agroindustry (sugar, cacao, rice, fruit,

vegetable, oil-palm, sweet-potato, meat, fish, etc) New Genetic Algorithms for Solving Multiobjective

Problems (S3) Methodology development (!)

51

Parallel Genetic Algorithms (S2, S3) Parallel and distributed computing New meta-heuristic techniques Super computer (multiple processor)

Bioinformatics (S2,S3) Application of information technology in molecular

biology, nuclear physics, chemical reaction, etc Computer Security (S1, S2, S3)

--- end of presentation ---