Because fuzzy sets exist deficiency on semantic description

much of the current research interest in neuro-fuzzy hybrid systems is focused on how to extend fuzzy neural networks. To deal with this problem

a self-organizing intuitionistic fuzzy networks based on UKF is presented. Firstly

structure of intuitionistic fuzzy networks and meanings of each layer is proposed. Secondly

training algorithm is deduced

and LLS and UKF are used to learn linear and non-linear parameters respectively. Thirdly

guideline of how to generate a new rule is given

and method of error descending rate is used as fuzzy rule pruning strategy

so that rule which plays an unimportant role in the system is deleted. At last

typical experiments of function approximation

system identification and prediction of time-series indicate that a fuzzy network obtained by the proposed algorithm has a more tighten structure and better generalization than other algorithms.