Forecasting Approach For Urban Function Nodes Based On Markov Chain And Neural Network

In the field of architecture generative design,it is often necessary to convert the architectural problems into mathematical models.However,such architectural problems which can be directly converted to mathematical models occupy only a small part of the complex architecture field,and there are still a large number of abstract problems that cannot be described by simple mathematical models.Therefore,this study regards the functional node prediction problem as the research objective which is a kind of those abstract problem,and uses the large amount of urban data information which is collected before as the data support,and explores some mathematical models to describe the abstract relationship between urban functional nodes.In the field of architecture generative design,a solution for the problem of mathematical modeling of the above abstract architectural problems can be provided after the study.Finally,this mathematical model can be used to solve the architectural problems which are related to urban functional nodes.The urban data collection and filter are used as the pre-order preparation and the Markov chain and neural network are used as description models respectively in the following research.The Markov chain is construced by the discrete event state definition and the discrete event state statistical division.For the neural network,the construction process is combined with training sample definition and model training.Finally,computer programming is used as a means of realization,and a mathematical model that can describe the relationship between functional nodes is obtained and used for function nodes prediction experiment and results discussion.The specific experimental results show that the Markov chain can describe the topological relations between functional nodes,the neural network can describe the positional and functional relationship between functional nodes,and these models have the ability to predict functional nodes under defined conditions.