Extended abstract
Background and purpose: Cities are like living and ever-changing organisms, they have a complex geometry, and the urban society is faced with newer and more complex issues day by day. Traditional urban design processes lack the necessary flexibility to deal with urban complexities. Therefore, to produce designs with high flexibility and responsiveness, design methods are needed that can produce different solutions and make the designer aware of the consequences of decisions at each stage and create a cycle of feedback. In this case, design will be a system of solutions for a problem.
Recent developments in urban societies such as epidemics, wars, economic disturbances, lack of access and shortage of sustainable and renewable resources and many other factors are the challenges that cities always face. In this regard, urban planners and designers can take effective steps to achieve sustainable cities by benefiting from past experiences and new tools. In this regard, it is important to use the artificial neural network algorithm in the simulation of sustainable urban design parameters, as one of the most widely used artificial intelligence algorithms, due to its ability to calculate quickly and with high accuracy based on the observations of software simulation.
This issue becomes important in problems where parametric simulation is possible, but the calculation time is very long, and the calculated function of the artificial neural network can perform the desired calculations with high accuracy and in a short time.
Investigation method: The present research is a developmental research in the context of quantitative-qualitative analysis, because by using the theoretical foundations of the subject, it studies the improvement of the quality of the environment in the formulation of the parametric urban design model and finds the optimal solutions in the context of intelligence. provides artificial In this applied-developmental research, using sustainable development indicators collected from various sources in cultural-social, institutional-management, environmental, economic and physical dimensions and using new tools of artificial intelligence, the field of urban design model compilation Parametric stability is provided.
In this way, the indicators of the sustainable city have been calculated by systematic analysis method. Then, the aforementioned indicators have been parametrically evaluated in the category of qualities expected from all types of urban spaces by content analysis method. In this regard, in order to measure each of the variables, a number of items and questions have been compiled, and the total of additional items and questions form the structure of the studies in the context of the interview, based on which the urban design variables have been parametric.
The way to answer the questions and issues has been to refer to the experts in the field of urban design, and in the next step, artificial intelligence, through learning from past experiences and mathematical calculations, will model urban spaces on the basis of the artificial neural network (ANN) structure as described; Pattern classification, clustering, function estimation, prediction, optimization, associative memory and control in management have been discussed.
Findings: The results obtained from the definition and explanation of the proposed algorithm show that in this way artificial intelligence accelerates the way to reach the goals by appropriate action and compatible with the complexities of the environment and through learning from past experiences and mathematical calculations. , even in the lack of knowledge and insufficient information resources, it has the ability to maximize the success of solving problems, which is one of the most important goals of research.
In this regard, first, the training data set is prepared through rule-based processes, and then, using the data-driven process, three artificial neural networks are trained and three separate fitting functions are approximated for them. These functions are set based on three different structures of artificial neural networks. At the end, by comparing the error rate of each model, the predicted values are also examined and the influence of the neural network structure in the obtained answers is analyzed.
Discussion and conclusion: based on the results of the research, the model presented in this research can be generalized to other similar problems of predicting the parameters of sustainable urban design, and there are two basic applications for these models. First, it creates the possibility of accurate calculation without the need for heavy calculations, and secondly, in complex simulations, by having independent and dependent parameters and without the need to study the governing rules, it has the ability to estimate the function. can predict the desired stability parameter with high accuracy.

Keywords: artificial intelligence, urban design, parametric urban design, urban context.