Testing of inflatable-structure shape control using genetic algorithms and neural networks

The active shape control of inflatable structures was analyzed with the help of genetic algorithms and neural network. The technique made use of a 200 x 300 mm rectangular Kapton membrane and a local thermal load source. The membrane flatness was found to be dependent on the thermal load and the tension combinations. The tension combinations were adjusted with the help of shape memory alloy wire actuators and tension values were obtained with the help of strain gages. A vision system was developed to measure the membrane flatness. The control system was implemented with the help of LabView, Matlab, and Automation Manager. LabView code regulated input parameters, results display, and data I/Q, while Automation Manager realized the vision system. The membrane-flatness control system was tested by using local heater. Results show the effectiveness of neural network and genetic algorithms in shape estimation of inflatable structure.