Learning soft computing control strategies in a modular neural network architecture

Modelling and control of nonlinear dynamical systems is a challenging problem since the dynamics of such systems change over their parameter space. Conventional methodologies for designing nonlinear control laws, such as gain scheduling, are effective because the designer partitions the overall complex control into a number of simpler sub-tasks. This paper describes a new genetic algorithm based method for the design of a modular neural network (MNN) control architecture that learns such partitions of an overall complex control task. Here a chromosome represents both the structure and parameters of an individual neural network in the MNN controller and a hierarchical fuzzy approach is used to select the chromosomes required to accomplish a given control task. This new strategy is applied to the end-point tracking of a single-link flexible manipulator modelled from experimental data. Results show that the MNN controller is simple to design and produces superior performance compared to a single neural network (SNN) controller which is theoretically capable of achieving the desired trajectory. (C) 2003 Elsevier Ltd. All rights reserved.

Textile wastewater treatment using granular activated carbon adsorption in fixed beds

This work involved the treatment of industrial wastewater from a nylon carpet printing plant which currently receives no treatment and is discharged to sea. As nylon is particularly difficult to dye, acid dyes are required for successful coloration and cause major problems with the plant's effluent disposal in terms of color removal. Granular activated carbon Filtrasorb 400 was used to treat a ternary solution of acid dyes and the process plant effluent containing the dyes in a fixed-bed column system. Experimental data were correlated using the bed depth service time (BDST) model to previously published work by the authors for single dye adsorption. The results were expressed in terms of the BDST adsorption capacity, in milligrams of adsorbate per gram of adsorbent, and indicated that there was a 12-25% decrease iri adsorption capacity in the ternary system compared to the single component system; This reduction has been attributed to competitive adsorption occurring in the ternary component system. Dye adsorption from the process plant effluent showed an approximate 65% decrease in adsorption capacity compared to the ternary solution system. This has been attributed to interference caused by the other colorless textile effluent pollutants found in the process wastewater. A chemical oxygen demand analysis on these components indicated that the dyes accounted for only 14% of the total oxygen demand.

Granular fertilizer drying and agglomeration in storage

In the manufacture of granular NPK fertilizer the product is cooled before packaging and storage in moisture-proof bags. It has been shown that the temperature of the fertilizer prior to packing is significant in that at high temperatures, drying of the granules takes place in the bag which causes an increase in the humidity of the air surrounding the granules and thus an increase in moisture content at the granule - granule interface. This surface moisture was shown to increase the likelihood of agglomeration in the fertilizer by a capillary adhesion/unconfined yield stress model. An iterative model was set up to establish conditions that would prevent drying occurring, which takes into account fertilizer drying rate, fertilizer cooling rate cooling rate and the effect of coating oils on the drying mechanism.