The process of predicting the failure of structures is a very complex and difficult process due to the different nature and density of the materials used in the buildings. In this study, the focus was on high-rise buildings constructed of steel, where the types of high-rise buildings constructed of steel were studied, as well as the systems adopted in the construction of these buildings in several countries around the world, which were as follows: classification is provided for structural systems for all types of tall structures i.e., steel structures, concrete structures and composite structures:1- Hard frame systems 2- Bracing frame systems and shear wall frame 3- Truss systems 4- Tubular frame systems 5- Bracing tube systems 6- Batch pipe systems. A number of examples of the failure of taller buildings that were subjected to earthquakes were also reviewed. This thesis was created to predict mainly the failure of high-rise steel structures in the capital, Tehran, in the Islamic Republic of Iran, for two reasons, given the nature of this geographical city located on the seismic line, and thus all buildings in this city are at risk of failure due to earthquakes. The other reason is the economic embargo imposed on the Islamic Republic by the United States of America, which in turn affects the economy and the financial abundance of the country. In view of what was mentioned above about the difference in the systems for the construction of tall buildings and the difference in the materials used in construction, and also Several types of artificial neural networks ANN have been studied to choose the best in order to be combined with the analytic hierarchy process AHP method. in this thesis the factors affecting buildings in the construction phase of the building were studied to predict the failure of the building in the future, which is divided in to four indicators and each indicators containing several factors. the following factors that have been effective in intelligently identifying the failure of high-rise steel structures using artificial intelligence methods and neural networks were identified as follows: 1-Financial Indicators: it includes four factors 2-Human Indicators: it includes four factors 3-Natural Indicators: it includes three factors 4-Technology Indicators: it includes three factors These factors and indicators have been studied and know the extent of their relationship to each other, as well as their relationship to the failure of structures, as it is found there is a relationship between the failure of structures and these factors, because these factors determine the extent of the efficiency of the facility to withstand natural disasters in the future Through making the questionnaire and VIII Abstract obtaining advice from specialists and engineers, a matrix of weights was did for the four indicators mentioned above. The analytic hierarchy process (AHP) was used to calculate the weight for each indicators and its impact on determining the prediction of the failure of structures, knowing that this matrix is not fixed, as it may change with the change of the area in which the questionnaire worked Or based on the opinion of experts in this field(it was did in Tehran city) It was concluded by AHP method that the financial indicators have the most influence with a percentage of 36%, followed by the human indicators by 28%, then the natural indicators by 20%, and finally the technological indicators by 16%. As can be seen from the results, among financial indicators, factor(1) as the most important factor and factor(3) as the least effective factor.As can be seen from the results, among human indicators, the factor(2) as the most important factor and factor(1) were obtained as the least effective factor. As can be seen from the results, among natural indicators, the factor(1) as the most important factor and the factor(3) as the least effective factor was obtained. As can be seen from the results, among technology indicators, the factor(2) as the most important factor and the factor(3) as the least effective factor was obtained. Priority samples were obtained for four structures and they were compared by means of a weights matrix to find out which are the most efficient (the least likely to fail) and the least efficient (the most likely to fail). Samples were taken including the priority of 200 buildings so that we can use them in artificial neural networks ANN, where the buildings were classified into two classes according to the final value of the priority, where the building whose have final value of priority is greater or equal to 0.005 is successful or rather less likely to fail, and less than 0.005 is failure or rather more prone to failure. This type of artificial neural network was used with these specifications to obtain the best classification of samples. Classifier model name: MLP Multilayers perceptron with backpropagation was used in this thesis. A special code was used to make this artificial neural network, and the classification results (test) for the last 20 samples out of a total of 200 with accuracy were very high, only one of the 20 was misclassified, knowing that the value was very close to the value 0.005 which represents the cut-off value in the final classification where the value was 0.004936 It is very close to the above value
