RUBBER APPLICATIONS DEMAND VULCANIZED PRODUCTS DUE TO IMPROVED PROPERTIES OF CROSSLINKED ELASTOMERS.SULFUR IS THE MOST WIDELY USED CROSSLINKING AGENT AND IS OFTEN ACCOMPANIED BY ACCELERATORS, RETARDERS, ACTIVATORS AND OTHER COMPONENTS.THREE MAIN REGIONS OF CURE CAN BE DISTINGUISHED IN THE VULCANIZATION (CURE) CURVE ARE INDUCTION, CURING, AND OVERCURE [1].INSTEAD OF THE PHENOMENOLOGICAL APPROACH, A KINETIC APPROACH IS USED TO SIMULATE THE ACCELERATED SULFUR VULCANIZATION. A COMPLEX MODEL REPRESENTATION OF THE REACTION WILL RESULT IN A SET OF NONLINEAR DIFFERENTIAL EQUATIONS. APPROPRIATE ASSUMPTIONS AND SIMPLIFICATIONS ARE USUALLY MADE TO ACHIEVE AN EASY NUMERICAL SOLUTION IN MOST PRACTICAL CASES. KINETIC PARAMETERS ARE OFTEN OBTAINED THROUGH A NONLINEAR CURVE FITTING ROUTINE [2].GENETIC ALGORITHMS (GA) BELONG TO THE FIELD OF EVOLUTIONARY ALGORITHMS WHOSE MAIN IDEA IS TO SIMULATE MECHANISMS DERIVED FROM BIOLOGY WITH AN AIM TO MODELING EVOLUTION. THE INITIAL POPULATION IS RANDOMLY CHOSEN IN THIS SPACE. THE INDIVIDUALS/SOLUTIONS ARE EVALUATED AND RANKED USING A FITNESS FUNCTION WHICH CAN ALSO BE THE OPTIMIZATION CRITERION. THEN, THE POPULATION PROGRESSIVELY MOVES TOWARDS FITTER REGIONS OF THE SEARCH SPACE BY MEANS OF BIOLOGICAL OPERATORS (CROSSOVER AND MUTATION.) [3].IN THIS WORK FOR SEVERAL FORMULATION OF RUBBER COMPOUND WERE PREPARED BY MILL MIXING AND ITS CURE CURVES WERE OBTAINED BY ODR RHEOMETER. THE SIMULATION DATA WERE DERIVED BY TWO MODEL OF REACTION [4, 5].THE EVALUATION OF KINETIC CONSTANTS FOR THE PROBLEM AT WERE ESTIMATED WITH GA.THE AGREEMENT WITH THE PRESENT APPROACH AND EXPERIMENTAL RESULTS SEEMS RATHER PROMISING, ALSO IN CONSIDERATION THAT VERY FEW EXPERIMENTAL POINTS ARE NEEDED AND TIME REQUIRED TO SOLVE WITH A GA WAS LEES THAN TIMES THAT REQUIRED BY A NON LINEAR PROGRAMMING ROUTINE.

VOLATILE ORGANIC COMPOUNDS (VOCS) PLAY AN IMPORTANT ROLE IN ATMOSPHERIC CHEMISTRY [1]. PROTONIUM IONS ARE PERFORMING NON-DISSOCIATIVE PROTON TRANSFER TO MOST OF COMMON VOLATILE COMPOUNDS BUT DO NOT REACT WITH ANY OF COMPONENT PRESENT IN CLEAN AIR. ALTHOUGH VARIOUS ANALYTICAL METHODOLOGIES HAVE BEEN DEVELOPED FOR MEASUREMENTS OF VOCS, THE APPROACH, WHICH BASE ON DETERMINING PROTON TRANSFER REACTION RATE CONSTANTS BETWEEN H3O+ IONS AND VOCS APPEARS TO OFFER GREATEST ADVANTAGE FOR DETECTING ORGANIC SPECIES [2]. UNDER THESE CONDITIONS, NUMEROUS CALCULATION PROCEDURES HAVE BEEN PROPOSED FOR ESTIMATING THIS PARAMETER. THE MOST COMMON APPROACH FOR ESTIMATING RATE CONSTANTS IS QUANTITATIVE STRUCTURE–PROPERTY RELATIONSHIP (QSPR) APPROACHES [3]. IN QSPR MODELS, THE STRUCTURAL PARAMETERS OF INTERESTED COMPOUNDS WERE CORRELATED TO THEIR PROPERTIES BY THE MEANS OF A MATHEMATICAL MODEL.IN THIS WORK A QSPR MODEL WAS DEVELOPED TO CORRELATE RELEVANT STRUCTURAL PARAMETERS OF 50 VOCS TO THEIR ION-MOLECULE RATE CONSTANT. MULTIPLE LINEAR REGRESSION (MLR), GENETIC ALGORITHM (GA) AND ARTIFICIAL NATURAL NETWORK (ANN) WAS USED TO DEVELOPING THESE MODELS [4]. THE MOST SIGNIFICANT DESCRIPTORS WHICH SELECTED BY STEPWISE MULTIPLE LINEAR REGRESSION WERE: MAXIMAL ELECTROTOPOLOGICAL POSITIVE VARIATION (MAXDP), D COMMA 2 VALUE/WEIGHTED BY ATOMIC VAN DER WAALS VOLUMES (DISPV), LEVERAGE-WEIGHTED AUTOCORRELATION OF LAG 5 /WEIGHTED BY ATOMIC MASSES (HATS5M), CH3R/CH4 (C-001) , AND RELATIVE NEGATIVE CHARGED(RNCS). THESE DESCRIPTORS WERE USED TO DEVELOPING SOME LINEAR AND NON-LINEAR MODELS. COMPARISON BETWEEN THE STATISTICAL PARAMETERS OF THESE MODELS REVEALS THE SUPERIORITY OF NONLINEAR MODEL OVER LINEAR MODEL. THE STATISTICS OF THE ANN MODEL WERE RTRAIN=0.986 SETRAIN=0.147, RINT=0.930 SEINT=0.285, REXT=0.900 SEEXT=0.427. ALSO THE DEVELOPED MODEL WAS FURTHER EVALUATE BY CROSS-VALIDATION TEST WHICH PRODUCES THE STATISTICS OF Q2=0.978 AND SPRESS=0.392. THE RESULTS OF THIS STUDY INDICATE THE APPLICABILITY OF ANN MODEL TO ESTIMATE THE PROTON TRANSFER RATE CONSTANTS OF VOLATILE ORGANIC COMPOUNDS.