The extraction of resinoids from St. John's wort (Hypericum perforatum L) was studied in a series of two papers. In the first part, the effects of the operating conditions on the yield of resinoids (total extract) were analyzed, while the mathematical models of extraction kinetics were compared in the second one. The extraction was carried out using an aqueous solution of ethanol (70 and 95 % v/v) at a hydromodulus (plant material to solvent ratio, w/v) of 1:5 or 1:10. The plant material was disintegrated and divided into three fractions (mean particle size: 0.23, 0.57 and 1.05 mm). The temperature was 25, 50 or about 80°C (boiling temperature). A higher yield of resinoids was obtained when the plant material of greater disintegration degree (0.23 mm) was treated with 70% v/v aqueous ethanol solution at higher hydromoduli (1:10) and temperatures (80°C). The effects of the operating factors on the yield of resinoids were estimated by using both the full factorial experimental plan 24 and artificial neuronic networks (ANN) of 3-4-1 topology. Of the two methods, the ANN one was found to be advantageous because of its capability of estimating the yield of resinoids in the whole range of the applied operating conditions
7'ke extended Kalmanfilter has been successfully applied to the feedforward and the recurrent neural network training. Recently intraiuced derivative-Jke filters (Unscented Kalman Filter and Divided Direerence Filter) ourper/orm the extended Kalman filter in nonlinear state estimation. In the parameter estimation of the feedforward neural nehvorks UKF and DDF are comparable or slightly better than EKF, with a sign@cant odvontage thut they do not demand calculation of the neural network Jacobian. In this paper, we consider the application of E m , UKF and DDF to the recurrent neural network training. The claw of non-linear outoregressive recurrent neural networks with exogenous inputs is chosen as a bmic architecture due IO its powerful representational capabilitiesis the vector of *The work of B. TodoravlC was supported by a Scholarship ofthe German Academic Exchange Service (DAAD) under the Stability Pact far South East Europe.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.