In this paper the method of representing chemical structures as arguments multidimensional functions ternary logic is considered. Computer experiment showed promising implementation of the proposed method ternary logic. A graphical representation of the results facilitates reverse synthesis of chemical compounds.

A number of the 196 organic compounds of different chemical classes with the known values of the maximum permissible concentrations established by sanitarytoxicological endpoint hazard in the water objects domestic, drinking, and culturalhousehold purpose. Identified fragments of molecules, which have an effect on the increase and reduction of Toxicological properties. Formed ten models of theoretical evaluation of the level of water pollution chemicals. The results obtained can be used for preliminary assessment of the risk of organic substances and identification of potentially hazardous pollutants of the environment.

Based on methods of the theory of pattern recognition with the use of a computer system SARD-21 identified areas of molecular modifications in order to obtain fungistatic activity of aromatic and heterocyclic compounds. Calculated hypothetical reference fungicidal activity and toxicity is formed. Quantitative criteria for structural similarity of the compounds with calculated standards activity is determined. The basic molecules used to modify the structure and functional elements, the most favorable for the modification in the fungicidal activity are identified.

We have formed a sequence of steps for evaluating properties of chemical compounds. Our experiments have proved that complex-valued perceptrons can be used for predicting chemical properties. An algorithm has been developed to additionally train complex-valued perceptrons.

The possibility of a theoretical estimate of the selective properties of organic solvents on the basis of joint use of the computer system SARD-21 and a complex-perceptron is investigated. The clustering and forecasting approach based on the selective properties of complex neural networks have been proposed. The results showed the advantage of complex application of two methods to determine the optimal directions of search for new effective solvents.

A theoretical approach to definition of hazard classes of new organic substances using computer system «SARD-21» are proposed. Predictive models for complex formed of interval estimation of the safe exposure levels of organic compounds in the water of reservoirs fish industry are announced. Complex of models tested at a number of organic compounds with known values of maximum allowable concentrations. Reliable results forecast (70% more) confirm the applicability of the proposed complex at the stage of preliminary toxicity assessment of organic substances.

На основе методов теории распознавания образов с применением компьютерной системыSARD21 определены направления молекулярной модификации с целью получения фунгицидноактивных ароматических и гетероциклических соединений. Сформированы расчетныегипотетические эталоны фунгицидной активности и токсичности. Определены количественныекритерии структурного сходства исследуемыхсоединений с рассчитанными эталонами активности. Выявлены базовые молекулы, используемые для модификации и функциональные элементы строения, наиболее благоприятные длямодификации в рамках фунгицидной активности.

The investigation is carried out and the approach to before experimental estimation and modeling of selective solvents for the extraction of aromatic hydrocarbons from petroleum products on the basis of patterns of communication «structure–property» within the computer system SARD-21 is offered. The regularities of the influence of the functional groups and their combinations on the selectivity of the solvent for the system «hexane–benzene» are revealed. Models to assess the level of selectivity of solvents of aromatic hydrocarbons have been developed. The database for the development of areas of modeling and synthesis of selective solvents is generated.

An approach to the before experimental estimation of pesticide properties of organic compounds on the basis of application software systems SARD-21 and a complex-perceptron is offered. The possibility of clustering and forecasting based on the fungicidal properties of complex neural networks is researched. The results show the advantage of combined application of two methods that can produce the best recommendations for the synthesis of new fungicidal compounds.

The influence of the chemical fragments of aromatic and heterocyclic compounds on the fungicidal activity and complex pesticide properties. Identified more than 400 combinations of functional groups that increase the fungicidal activity. The common chemical fragments characteristic of fungicide and herbicide, fungicide and insecticide active. The results obtained can be used to develop new pesticides multifunctional purpose.

On the base of 170 organic structurally diverse compounds investigated dependence «Structure– toxicity» and possibility assessment of maximum allowable concentration toxic pollutants in the air on the basis of «SARD-21» (Structure Activity Relationship & Design). A result of research identified chemical fragments that increase or decrease the level of toxic hazard index. These findings can be used at modeling and before experimental estimate of maximum allowable concentration organic toxic pollutants in the air.

We consider two ways of identification of structures of chemical compounds—software SARD-21 and the complex-valued perceptron. Results of computational experiments showed the advantage of a neural network. In addition the complex-valued perceptron can produce meaningful recommendations for the synthesis of new compounds.

Study of 176 organic compounds of different chemical classes with well-known value of the defined admissible concentration in the water of water objects domestic, drinking, and cultural and community purpose, established by sanitary-toxicological endpoint hazard. On the basis of the theory of pattern recognition with the use of the computer system of the SARD-21 identified chemical fragments of structures, which have an effect on the increase and reduction of toxicological properties, and also formed a model of evaluation intervals of approximately permissible levels water pollution by organic substances. The results obtained can be used for the assessment of anthropogenic impacts of chemicals on wildlife and identification of potentially hazardous substances.