Abstract: A new modeling method is proposed to establish the soft sensor model by the analysis of several fuzzy modeling methods.Based on the idea of variable structure,the new method designs the initial structure model uniformly,and adds new fuzzy rules to improve the precision of T-S fuzzy system using the maximum absolute error(MAE)as the index.In order to accelerate the computational convergence of the parameter optimization,the model parameter optimization depends partly on the Levenberg Marquardt method for nonlinear parameter optimization and partly on the least square method for linear parameter estimation.Examples show that the proposed method makes the rules distributing rationally,computational convergence fast,and high generalization capability is got.The effectiveness of the method is illustrated by a number of multi input soft sensor simulation examples.