Research On Interval Information Granule Modeling Method Based On TS System

As an important research direction in the field of data analysis,regression model has been widely used in medical,aerospace,industrial production,and other fields.With the increasing complexity of the research object,in order to describe the characteristics of the system more comprehensively,many scholars introduce the idea of granular computing into the process of system modeling,and construct models that output different forms of information granules.The corresponding models can not only fit the modeling data,but also provide semantic interpretation for the modeling results.In this paper,interval information granulation technology is applied to fuzzy Takagi-Sugeno(TS)system modeling and interpretability analysis.The main contents of this paper are as follows.Firstly,this paper proposes a direct granularity model based on TS fuzzy rules.Compared with the traditional granularity model,this model does not need to build a numerical model in advance.Instead,it extracts the membership degree of antecedent and the output of consequent from the rules of fuzzy system to form weighted data,and then generates information granules through weighted data.Direct granularity model does not increase the number of model parameters in the process of building,which reduces the complexity of the model.Then,this paper establishes an improved performance index of the granular model,which can reflect the distance deviation between the output information granule and the target value,making the training process of the model more efficient.On this basis,an interval granular model with adaptive specificity weight is designed.Numerical simulation results show that the designed model can effectively cover the numerical sample information.Secondly,the TS fuzzy system is applied to the feature extraction of the broad learning system.By clustering the input data,a variety of features of the data are extracted,and then an improved fuzzy broad learning model is constructed.Then,a granularity allocation module is added to the output space,and a fuzzy broad learning system model based on granularity output space is established.For granular data,this paper also designs a fuzzy broad learning system which is trained directly by granular data.Through the numerical simulation of different types of data sets,the test results illustrate the effectiveness of the proposed method.