ARTIFICIAL NEURAL NETWORK MODEL FOR FLEXURAL DESIGN OF CONCRETE HYDRAULIC STRUCTURES

As a computer technique, Artificial Neural Networks (ANNs) have expanded in use with engineering fields. ANNs have been used in many civil engineering problems and some of them were used in the design of concrete structural elements and have shown a good degree of success. This paper presents an ANN model for the strength design of reinforced-concrete hydraulic structure according to the requirements of the Engineering Manual 1110-2-2104. Structural design is a sequential process needs iteration, assumption, checking the limits, ...etc, and that can be programmed with some judgments of the designer. 288 cases of design samples have been calculated using excel sheets and Microsoft visual basic programming language. 200 samples of design randomly selected have been used for training of (4-10-10-2) ANN model. 50 samples have been selected for validation, and 38 for prediction processes. The predicted design outputs were the thickness of hydraulic concrete section and corresponding steel reinforcements. Visual Gene Developer software of ANN prediction for general purposes has been used with a feed-forward neural network with a standard back-propagation learning process. The suggested artificial neural network model has predicted the output data of design for concrete sections, and the results have shown a satisfactorily match with the actual output data of design