Mark Davidovich KATS, born in 1937, higher education, specialist in applied mathematics, Sc.D., academician of the Ukrainian Technological Academy and Ukrainian Academy of Ecology, corresponding member of International Academy of Computer Sciences and Systems.

Home address: Ukraine, Lugansk region, 349940 Severodonetsk, Donetskaya street, build.37, apart.24, Tel: (380-6452) 30875, E-mail: root@ixt.lg.ua (Kats M.D.).

2. Project name.

"Optimization of operating production plants based on data recorded during normal operation".

3. Application field.

Operating production plants in chemical, petrochemical, biotechnological, metallurgical, chemical and pharmaceutical and other industries.

4. Proposal Description.

Operating plant performance analysis showed that most of them, especially those involving periodic (cyclic) processes in chemistry, metallurgy, biotechnology etc., are not operating in optimum conditions and their economic and environmental figures can be significantly improved.

It is well known that optimization of complex process operations is possible only when it is based on a sound mathematical model, suitable for optimum management synthesis.

However at present we don’t have available adequate mathematical tools which would allow to identify (to build mathematical models based on experimental data) multiple tasks related to study (mathematical modeling) and improvement (optimization) of existing technological processes referred to the class of "big" systems based on their informational characteristics (high dimensions of input and output variable vectors; complicated non-linear dependencies of input and output variables; inconsistency in output variables - optimization of one variable results in worse figures for other variables, etc.).

IN COLLABORATION WITH PRODUCERS OF CHEMICAL, PETROCHEMICAL, BIOTECHNOLOGICAL, PHARMACEUTICAL AND OTHER PRODUCTS WE WILL DEVELOP RECOMMENDATIONS USING NEW IDENTIFICATION AND SUBOPTIMIZATION METHODS TO IMPROVE PERFORMANCE OF OPERATING PRODUCTION PLANTS BASED ON CUSTOMERS’ CRITERIA OR COMPLEXES OF CRITERIA. SUBOPTIMIZATION RECOMMENDATIONS PRESENTED AS APPROPRIATE CHANGES OF EXISTING PROCESS RATES CAN BE IMPLEMENTED USING EXISTING INFORMATION-SUPPORT AND CONTROL SYSTEMS, SO THEIR REALIZATION WON’T REQUIRE ADDITIONAL INVESTMENT.

5. Advantages over analogs.

Currently no identification and optimization methods are known which would enable to significantly improve efficiency of complex process operations based only on experimental information about their input (ratio and quality of raw material feeds, process rates etc.) and output (economic, environmental and technological indices of the process) variables.

OPTIMIZATION OF MORE THAN 100 OPERATING PRODUCTION PLANTS SHOWS THAT OPTIMIZATION COST PAYBACK PERIOD DOESN’T EXCEED THREE MONTHS. AND IN SOME INSTANCES, AS WAS THE CASE WITH BELGOROD VITAMIN PLANT, PRODUCING B1 VITAMIN, AND BEREZNIKI CLEVE’S ACID PLANT, THIS PERIOD DID NOT EXCEED TWO WEEKS.

6. Potential users.

These include companies producing chemical, petrochemical, biotechnological, pharmaceutical and other products.

7. Extent of finalization.

RESULTING FROM STRATEGIC RESEARCH AIMED AT FINDING NEW APPROACHES WHICH WOULD PERMIT TO SOLVE PROBLEM OF CORRECT STUDY AND IMPROVEMENT OF COMPLEX TARGETS, WHICH BY THEIR INFORMATIONAL CHARACTERISTICS ARE REFERRED TO "BIG" SYSTEMS, BASED ON EXPERIMENTAL DATA GAINED IN MONITORING MODE (using usual methods of mathematical modeling and optimization to solve these tasks leads to serious and in most cases insoluble methodological and computing problems)

THERE WERE DEVELOPED METHODS OF MOSAIC PORTRAIT AND LOGIC PROGRAMMING BASED ON IDEAS OF ARTIFICIAL INTELLIGENCE, WHICH PASSED TESTING AND MANY YEARS EXPERIMENTAL VERIFICATION HAVING BEEN USED FOR SOLVING MULTIPLE SCIENTIFIC, RESEARCH AND APPLIED TASKS.

MOSAIC PORTRAIT METHOD PERMITS USING EXPERIMENTAL DATA TABLE ONLY, WITHOUT PARTICIPATION OF AN EXPERT IN THE RELEVANT FIELD OF KNOWLEDGE , TO BUILD COMPREHENSIVE MATHEMATICAL MODELS OF AN OBJECT OF ANY PHYSICAL NATURE, CONTAINING LARGE VOLUME OF NEW, NON-TRIVIAL, EARLIER UNKNOWN INFORMATION ABOUT PRINCIPALS DETERMINING RELATIONS BETWEEN ITS INPUT AND OUTPUT VARIABLES.

Applied to existing processes this mosaic portrait method allows to build a mathematical model which would adequately portray dependence of economic, environmental and technological characteristics of a plant performance on raw material quality figures and process rates.

To suboptimize processes under investigation using their mosaic models there was developed logic programming method (discrete programming task) which would permit to synthesize process management recommendations (represented as relevant subranges for all controlled variables). Following these recommendations enables improvement of process performance on specified complex of output variables.

There are available computer programs for realization of mosaic portrait and logic programming methods as well as many years experience in optimizing operating cyclic processes in many industries.

8. Proposal protection.

Mosaic portrait and logic programming methods are not protected, but most algorithms providing polynomial of labor input, respectively, implementation time in task dimension were not published and computer programs enabling their realization are the author’s property.