acer fabrics are dual layer structure comprise of two separate outer and one middle layers. The monofilaments of middle layer connect the two outer layers together. Various fabric formation techniques can be used in manufacture of spacer fabrics. Incorporation of the middle layer not only imparts unique transfer properties to the fabric, but also enhances fabric impact absorption capacity. These unique properties have resulted in extensive application of spacer fabrics in various areas such as automotive, medical, technical and industrial textiles, geotextile, sports textiles and apparels. Additionally these fabrics due to their unique properties have replaced Polyurethane or Neoprn foams in various applications. Orientation of fibers both along the surface and thickness of spacer fabrics not only creates a three-dimensional structure but also imparts a very high degree of reversible compressibility. As far as properties such as heat treansfer , moisture transfer and air permeability are concerned, spacer fabrics are much superior in these respects in comparison to conventional flat textile fabrics. Upon exertion of pressure, the knitted loops constituting the fabric are deformed, the deforming energy is absorbed by the loop and is transferred to the adjacent loops. Upon removal of pressure the deformed loops return to their initial configuration. Spacer fabrics due to their three-dimensional structure, high porosity and air permeability and unique heat and moisture transfer ability can easily prevent moisture formation on the skin. These together with dynamic air circulation within the fabric constitute the requirements of fabric comfort which is of paramount importance in various applications of textiles. In this study a model using artificial neutral network technique was developed. The ideal network was found by manipulation of network topology, epoch and activation functions. This model predicts transfer and compression properties of the spacer fabrics using parameters such as fabric mass, composition, porosity and thickness as inputs. In order to train the network, specifications of 30 commercially available samples of spacer fabrics were used to train the model. The model was tested by comparison of the theoretical and experimentally obtained set of results for further five commercially available samples of spacer fabric. A discrepancy of less than 10% was found to exist between the two sets of results. This confirmed the validity of the developed ANN model
