Вы находитесь на странице: 1из 6

EngOpt 2008 - International Conference on Engineering Optimization

Rio de Janeiro, Brazil, 01 - 05 June 2008

High Technology Innovation Management Model using Fuzzy Logic


Artur Wrzalik , Leszek Kiełtyka, Robert Kucęba

Czestochowa University of Technology, Management Faculty, Department of Management Information Systems,


Czestochowa, Poland
email: arturw@zim.pcz.pl; lekiel@zim.pcz.pl; robertk@zim.pcz.pl

1. Abstract: The main goal of this paper is to present innovation venture evaluation methodology, HT and process
innovations in particular. The elaborated methodology uses fuzzy systems characteristics adapted in decision making
processes. The paper shows stages of decision making processes modeling in the researched scope. The authors present
sample model verifications.

2. Key words: HT innovations, fuzzy systems, decision making processes modeling.

3. Introduction

The idea of innovativeness is characterized by many approaches to the issue. The following ones can be
distinguished among them: economic, social, organizational and technical ones. It was P. Drucker who formulated one of the
most important from the management point of view definitions. He defined innovativeness as a specific tool of enterprise –
an activity that gives resources new possibilities of wealth creation. According to him, businessmen use innovativeness of
changes as a chance to start new economic activity or providing new services [2].
It can be stated then, that innovativeness is a means of enterprise development in all the areas of its functioning and influence
on environment.
In the present paper its authors concentrated on discussing elaborated HT Innovation Management Model in the context of
decision making processes. The HT Innovation Management Model was developed according to the following integral and
chronological schedule:
 Determining HT innovativeness measures based on analysis of previous documents of International Systems of
Innovativeness Monitoring;
 Multi-criteria selection of generated HT innovativeness measures as linguistic variables obtained on the basis of
questionnaire survey HT Quality Measures Evaluation conducted in the chosen expert group;
 Determining membership function of individual input variables (HT innovativeness measures);
 Generating decision making rules databases on the basis of experts’ mental and verbal knowledge;
 Adapting defined rules in the inference unit of the proposed model;
 Input variables fuzzification realized on the grounds of defined membership functions;
 Fuzzy reasoning and analysis of obtained output variables.

4. HT innovations measures selection

Generation of innovation measures sets was based on source analysis and review of previous documents of
International Systems of Innovativeness Monitoring. In order to do that and meet paper’s assumptions, there were HT
innovation measures selected in the global and regional dimension as well as on the level of production enterprise. They were
selected from process, subject, marketing and organizational innovation measures. Innovativeness evaluation measures were
selected on the basis of defined by European Union standards such as [1, 3, 4, 5]:
 Complex Community Innovation Survey (CIS);
 European Innovation Scoreboard (EIS);
 National Experts on Science and Technology Indicators (NESTI);
 Oslo Manual Methodology;
 Frascati Manual Methodology.
Selected measures were evaluated by the chosen expert group on the basis of HT Quality Measures Evaluation questionnaire.
To achieve this goal HT innovation measures evaluation criteria were defined. They include:
 Utilitarianism – possibility to implement global perspective of enterprise innovativeness in the measure;
 Compatibility – possibility to adapt innovation measures in various organization types, including network
organizations of global and regional dimension;
 Convergence – possibility to adapt HT innovation measures in various areas of production organizations activity;
 Comparativeness – applying uniform reference units for selected HT innovation measures group;
 Nature in microeconomic perspective – applying in the measure enterprise innovativeness in the perspective of
economic indexes of investment on innovations repayment;
 Nature in macroeconomic perspective – applying in the measure factors influencing economic growth of regions;
 Social-community nature – measuring education level with reference to technological culture and growing
awareness of the created IT society in global and regional perspective.

5. Structure of HT Innovations Management Model

The HT Innovations Management Model using fuzzy logic has been defined within the confines of the conducted
research. The proposed model represents mental knowledge of domain experts in the area of innovations management. The
sum of experts’ open and tactic knowledge was determined through mental model. Rules database allocated in the inference
.

unit of the linguistic model results from translating verbal model information. Initial premises constituting the basis for
defining aggregated set of rule were determined in the structure of the verbal model, which was created in accordance with
the instructions of the experts. The verbal model and the mental model constitute an inclusion of the proposed in the course
of the research fuzzy model [6].
Strict values undergo fuzzification in the fuzzification module of the proposed model. They represent relative and absolute
measures of features describing the phenomenon of HT innovation management. Analysis of information obtained from the
domain experts and study of previous documents allowed to define six input variables:
 Production processes automation resources (PPAR),
 Automatic production lines (APL),
 Production line computer control (PLCC),
 Industrial robots manipulators (IRM),
 Informatic systems of technological process management (ISTPM),
 New IT technologies (NITT).

Figure 1 presents chosen input variable creation (PPAR) in 16 regions of Poland. Presented values constitute percentage
of enterprises that implemented production processes automation resources within introduced technological innovations

Production processes automation resources [% ]

60
50
40
30
20
10
0
EG 1

RE N 2

RE N 3

EG 4

RE N 5

RE N 6

EG 7

RE N 8

10

EG 12

EG 14

16
RE N 1

RE N 1

RE N 1
N

RE ON

N
IO

IO

IO

IO

IO

IO

IO

IO

IO

IO

IO

IO

IO

IO

IO
I
G

G
RE

RE
R

Figure 1. Chosen input variable (PPAR) in percentage perspective


Source: on the basis [10]

On the basis of domain experts’ instructions input variables selected in the course of the research were described
with four fuzzy sets represented by trapeze membership functions (Figure 2). A similar nature of membership function was
determined for output variable describing innovativeness level of the region (IL). Just like in case of input variable four states
were defined for it – low, medium, high and advanced.

Figure 2. Membership function shape for the chosen input variable (IRM)
Source: own analysis

2
.

The research used MISO fuzzy model (many inputs – one output), which uses implemented method of fuzzy
concluding based on Mamdani’s model (Figure 3). Models of this kind treat the system that is being modeled as a black box
that is characterized by the fact that it lacks any information on physical phenomena occurring inside it. This is the result of
the fact that there is no unambiguous algorithm or mathematical model describing the formula of innovation level evaluation.
Mamadani model’s rules database determines strength of input- output relationships only. It does not define mathematic
interpretation of the analyzed phenomenon [7].
In the next stage fuzzy input variables underwent inference (fuzzy concluding). Heuristic rules database defined by the
domain experts was used here. It included the following premises that characterize fuzzy models’ functioning [8]:
 Evaluating all rules premises fulfillment level,
 Determining activated conclusion membership functions of individual rules’
 Determining output conclusion membership function of all rules.

The output membership function is calculated in the inference module of the proposed model. The calculation is based on
input membership levels and the output membership functions shows innovativeness level. The inference module structure
contains following elements:
 Rules database,
 Inference mechanism,
 Output membership function of the model.

Figure 3. Structure of HT Innovations Management Model


Source: own analysis

Sample mathematic notation of the chosen membership function for output variable (PPAR production processes automation
resources) looks as follows:



 0 for x < 34
 x − 34
 for 34 < x ≤ 36
2
f ( x) =  (1)
 1 for 36 < x ≤ 38
 x − 40
 for 38 < x ≤ 40
 2
 0 for x > 40

Where f(x) shows membership level of the set of regional enterprises where the usage level of production processes
automation resources is average.

3
.

Rules database development was based on information acquired from the verbal model. Logical rules that define cause-result
relationships among six input variables and the output variable defining innovativeness level were defined in this base
(Figure 4 and Figure 5).

Figure 4. Graphic interpretation of chosen rules defined in inference module


Source: own analysis

Figure 5. Input-output representation surface for ISTPM and APL input variables
Source: own analysis

Due to the nature of the membership function the defined by domain experts rules database contained 4096 rules.
They enabled unambiguous evaluation of chosen regions’ innovativeness level assigning them to particular output valuable
fuzzy sets. Sample database rules are shown in Table 1.

Table 1.Chosen rules allocated in inference module

Premises Conclusion

If (PPAR is low) and (APL is low) and (PLCC is average) and then
(IRM is high) and (ISTPM is low) and (NITT is low) (IL is low)
If (PPAR is average) and (APL is average) and (PLCC is high) then
and (IRM is low) and (ISTPM is advanced) and (NITT is average) (IL is average)

If (PPAR is low) and (APL is high) and (PLCC is high) and (IRM then
is low) and (ISTPM is average) and (NITT is low) (IL is average)

4
.

If (PPAR is advanced) and (APL is low) and (PLCC is low) and then
(IRM is high) and (ISTPM is average) and (NITT is average) (IL is average)

If (PPAR is high) and (APL is low) and (PLCC is high) and (IRM then
is average) and (ISTPM is low) and (NITT is advanced) (IL is high)

If (PPAR is low) and (APL is high) and (PLCC is advanced) and then
(IRM is high) and (ISTPM is average) and (NITT is average) (IL is high)

If (PPAR is average) and (APL is low) and (PLCC is high) and then
(IRM is high) and (ISTPM is advanced) and (NITT is advanced) (IL is high

If (PPAR is advanced) and (APL is advanced) and (PLCC is then


average) and (IRM is high) and (ISTPM is advanced) and (NITT (IL is high)
is average)

If (PPAR is low) and (APL is average) and (PLCC is advanced) then


and (IRM is advanced) and (ISTPM is advanced) and (NITT is (IL is high)
advanced)
If (PPAR is high) and (APL is advanced) and (PLCC is high) and then
(IRM is advanced) and (ISTPM is advanced) and (NITT is (IL is advanced )
advanced)
Source: own analysis

When designing rules database the authors tried to define the relationships between input and output variables in
such a way that completeness and non-contradiction criteria of this base were preserved.
The inference mechanism of the model realizes evaluation of individual rules premises fulfillment using t-norm and
s-norm operators (-min/prod for the conjunction premise, -max for the alternative premise). Next stage realized in the
inference module is calculating individual rules conclusion activation using Mamdani’s implication operator [9].
An aggregated set of modified conclusion functions is generated as a result of the inference in the individual rules of the rules
database. It serves as a basis to determine one final membership function of the whole rules database. The presented process
of final conclusion creation on the basis of components conclusion is called accumulation. Its realization is based on s-norm –
max operator.
The received through inference fuzzy value determined within fuzzy sets is then made more strict in the defuzzification
module. In the research the strictness mechanism used the centre of gravity method [7].
Additionally, the authors designed in their model a mechanism of received steady value translation into a linguistic form. It
presents region’s innovativeness level in the four-point scale: low, average, high and advanced.

6. Summary

The research shows that innovativeness level evaluation in the enterprises of the 16 regions of Poland is as follows:
 18% of the regions analyzed is characterized by low enterprise innovativeness level,
 In 26% of the regions enterprise innovativeness level is average,
 In 31% of the analyzed regions enterprise innovativeness level is high,
 25% of the chosen Polish regions are characterized by advanced enterprise innovativeness level.

The research results present innovativeness level evaluation in the chosen regions of Poland. The regions differ
geographically, socially and in the trade of the researched enterprises. It must be noticed here that both the nature of the
regions analyzed with reference to the modeled phenomenon as well as flexibility of fuzzy logic used in the research
enable application of the presented model in global and production enterprise perspective. The proposed HT Innovations
Management Model may serve as advisory one in case of decisions concerning innovations introduction. It may also
function as advisory model for evaluating already used innovations in order to assess if any updates are needed.
The analysis was conducted according to the assumptions of research project nr N N115 137 234

7. References

1. A. Arundel and H. Hollanders, 2006 “Global Innovation Scoreborad” (GIS) Report, MERIT – Maastricht
Economic and social Research and training centre on Innovation and Technology, European Trend Chart on
Innovation, European Commission, 2006,
2. P. Drucker, Effective manager. Warszawa: Wydawnictwo MT Biznes, 2007,
3. European Innovation Progress Report 2006, European Commission, ISNN 1683-349X;
http://www.trendchart.cordis.lu/Reports/Documents/EIPR2006-final.pdf,
4. European Innovation Scoreboard 2005. Comparative analysis of innovation performance. European Trend Chart on
Innovation, European Commission, 2005,
5. European Innovation Scoreboard 2006-Methodology. EIS 2006 Indicators; http://www.proinno-
europe.eu/index.cfm,

5
.

6. L. Kiełtyka and R. Kucęba and A. Sokołowski and A. Wrzalik, Knowledge manager’s decision ma king model on
the basis of fuzzy systems. [W:] Kiełtyka L.: Technologies and communication systems, knowledge and
information management. Redakcja naukowa: L. Kiełtyka. Warszawa: Wydawnictwo Difin, 2008.
7. A. Piegat, Fuzzy control and modeling. Warszawa: Wydawnictwo Exit, 1998,
8. L. Rutkowski, Artificial intelligence methods and techniques. Warszawa: Wydawnictwo Naukowe PWN, 2006,
9. A. Sokołowski, and A. Wrzalik and R. Kucęba and M. Pudło Rules database heuristics in linguistic model of
manager evaluation, [in:] IT economic organizations, red. L. Kiełtyka, Toruń: Dom Organizatora, 2007,
10. A. śołnierski [red.], Innovativeness 2006. State of innovativeness, support methods, research programmes - Report.
Warszawa: Polska Agencja Rozwoju Przedsiębiorczości, 2006;
http://www.pptb.pl/publikacje/raport_innowacjnosc.pdf,

Вам также может понравиться