摘要：The parameters and sometimes even the structure of the real complex systems are not exactly known in advance, and which moreover often change in unpredictable way. Adequacy of models of such complex systems must be then ensured by identification of their structure and parameters. One of suitable method of modelling of such ill-known and difficult measure systems appears a fuzzy non-linear regression analysis represets by Takagi-Sugeno fuzzy model. In the paper an extended TS model is presented with the regression coefficients in the shape of fuzzy numbers. The difference between the actual and computed values of the dependent variable in the new fuzzy regression models - with the fuzzy regression coefficients - are due to the "indefiniteness" of the system structure and parameters and - in the end - the level of system fuzziness is expressed through to fuzziness of model output variable.
