The application of neural networks in the selection of location datum

Selection of locating datum is the basis of the fixture planning, which has a direct influence on the quality of the clamping scheme and the machining quality of workpiece. Selection of locating datum is a complicated process. The designer considers not only the locating feasibility but as well surface type, valid locating area, surface roughness and tolerance relation, etc. In practice the design of a fixture relies heavily on the designer´s expertise and experience up to now. The paper determines 5 key factors influencing the selection of locating datum of workpiece firstly, the feature information of these influence factors are normalized by defuzzy reasoning. Finally the weights of the influence factors which are obtained according to experience are determined by the training of neural networks. The validity of the method is proved by a case.